Impact of Task, Structure, and Environment on Electronic Health Record Adoption, Use,
and Interoperability in Hospitals
A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL
OF THE UNIVERSITY OF MINNESOTA BY
Young-Taek Park, M.P.H.
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Stuart M. Speedie, PhD, Adviser
June 2010
© Young-Taek Park 2010
i
Acknowledgements
First of all, I would like to give my deep thanks to Dr. Stuart M. Speedie, my academic advisor. He allowed me to study health informatics and also financially supported me. I learned his teaching philosophy, knowledge, and research method during my study. Without his help and support, I might not have finished my dissertation. Thank you, Dr. Speedie, for all you have done for me. You will always be my respected teacher. I would also like to thank Dr. Laël C. Gatewood, the chair of my Ph.D. dissertation committee; Dr. Jeffrey S. McCullough, School of Public Health; and Dr. Martin LaVenture, Director of the Center for Health Informatics, Minnesota Department of Health, for serving on my Ph.D. dissertation committee. Although they had their own busy schedules, they willingly helped me complete my dissertation. My personal thanks should be given to Dr. Donald P. Connelly, faculty in the Health Informatics Graduate Program, who gave me a research opportunity and thereby provided financial support as well. I also would like to express my special gratitude to Dr. A. Marshall McBean and Dr. Bryan E. Dowd, Department of Health Policy and Management, School of Public Health. Dr. McBean taught me personally and academically and gave me lots of research opportunities analyzing Medicare claim data. Dr. Dowd helped to open my eyes in the field of health informatics. I owe these three so much more than I can pay back over the course of my life. I would also like to give my gratitude to the Health Informatics Graduate Faculty: Dr. Linda Ellis, Dr. Stanley Finkelstein, Dr. John Faughnan, Dr. Jeffrey Hertzberg, Dr. Bonnie Westra, Dr. Terrence Adam, Dr. Julie A. Jacko, Dr. James R. Fricton, Dr. Layne Johnson, and Dr. Genevieve Melton-Meaux. During my study, they taught me both directly and indirectly which contributed to the expansion of my academic knowledge of health informatics. I would also like to thank Chief Information Officers (CIOs) and IT Department managers in Korean hospitals who willingly participated in the survey. Especially, I thank Mr. Sung-Jeek Jung, director of the IT Department of Ilsan Hospital, the National Health Insurance Corporation. He introduced me to many CIOs, specifically the Korea Information Technology of Hospital Association (KITHA). I would like to thank the member hospitals of KITHA for their help through posting my research study into their main homepage website and voluntary participating in the survey. I also thank Dr. Young-Ho Oh and Dr. Na-Mi Hwang at the Korea Institute for Health and Social Affairs and Dr. Yong-Kyoon Lee at the Korean Institute of Hospital Management who always encouraged and supported me.
ii
I also thank the many fellow student colleagues who helped and supported me during my study, Dr. Nawanan Theera-Ampornpunt, Dr. Hojung Yoon, Mr. Joel Ackerman, Mr. Jeff Gao, Miss Kanaporn Tansriprapasiri, Mr. Stephen Yeboah, Ms. Resty Namata, and Miss SungRim Moon in the Health Informatics Graduate Program, and Mrs. Jing Du, Mr. Jinhyung Lee, and Mrs. Jeehae Kim in the Department of Health Policy and Management, School of Public Health. Much credit should be given to Miss Jessica Whitcomb-Trance at the Institute for Health Informatics (IHI) who corrected the grammar in my Ph.D. thesis manuscript. Christina M. Davenport, JD, at the University of St. Thomas School of Law also gave me good comments on my Ph.D. thesis. Mr. Wenjun Kang, staff at the IHI and Ph.D. candidate in the Department of Biostatics assisted me with downloading several important datasets for my thesis from websites located in South Korea. I also thank Mrs. Jiayan Huang, a M.S. graduate in Biostatistics in the School of Public Health, who gave me several statistical consultations. I would like to deep thank Dr. Han Joong Kim, the current President of the Yonsei University in Korea, who was my academic advisor in the Public Health Graduate Program and encouraged me throughout my studies. I also would like to thank Dr. Kyu Sik Lee and Dr. Eun Woo Nam, Department of Health Administration at the College of Health Sciences, Yonsei University, who always encouraged and supported me a lot. Dr. Young Moon Chae and Dr. Woojin Chung, Public Health Graduate Program of Yonsei University, also gave me good advice on the life of foreign studies. I also cannot forget much help from one of my closest friends, Dr. Jae Sung Park, Department of Health Care Administration at Kosin University in Korea. Lastly, I would like to thank my father-in law, Mr. Kwang-Hae Lim, and mother-in law, Mrs. Myung-Ja Park, who supported me a long time. I would like to thank my four sisters and two brothers for their endless love and encouragement, especially my two older sisters, Mrs. Kyoung-Sook Park and Mrs. Kyoung-Hee Park. In the U.S., I met a wonderful American family, Mrs. Colleen Diffely-Sullivan, Mr. Michael J. Sullivan, and their two children, Aaron and Ashley. I was not lonely because of them. I would also like to thank Dr. Harry Hull, Mrs. Suzi Henrichs, Mr. Ray Sandborgh, and Mrs. Anita Sandborgh who were always with my family. Lastly, I would like to express tremendous gratitude to my wife, Dongock Im, and two children, Sur-Hi, and Alexander Joonseo. Sorry for the long delay in my study and thank you so much for your solid support. I love all of you so much. Thank you, God, for giving me this great achievement.
Young-Taek Park June, 2010
iii
Dedication
I would like to dedicate this Ph.D. dissertation to my mother, Ock-Jin Kim, who is
always praying for my safety and well-being, and my father, Wee-Sun Park, who passed
away when I was a young child. My mother has put her everything into my education and
well-being since I became handicap, due to a big traffic accident, at the age of five. Sorry
Mother, for giving you so many worries and concerns. You don’t have to worry so much
about me anymore! I still feel my father’s deep love for me and keep several memories
that I had with him. Father, I remember you made a wooden kite reel for me. After you
left for heaven, it was broken. I could not fix it without you. Thank you, both of you, for
giving me the joy of life…
iv
Abstract
A paradigm in the field of Heath Informatics which has been taken for granted up until
this point may be disappearing and a new paradigm may begin to take shape as paper-
based medical record (PMR) systems are changing to the electronic health record (EHR)
systems. Although the PMR has played a critical role in recording patient’s clinical
information, now many studies report that EHR systems improve quality of care beyond
PMRs. For this reason, the governments across the world have initiated various
approaches accelerating EHR adoption, use, and interoperability. However, there has
been a paucity of studies explaining which factors affect EHR adoption, use, and
interoperability in hospitals. The objective of this study is to predict and investigate those
factors.
This study used a non-experimental, retrospective, cross-sectional study design to
measure hospitals’ internal features. Specifically, this study conducted a nationwide EHR
survey with the IT departments in South Korean hospitals by using online surveys from
April 10 to August 3, 2009. It used Generalized Estimating Equations, an extension of the
Generalized Linear Model, to interpret EHR system adoption and interoperability, and
General Linear Mixed Model for the use of EHR systems.
With respect to EHR system adoption, this study found that 1) the likelihood of EHR
adoption increases as a hospital’s task complexity - measured by the number of medical
specialties - IT infrastructure, and organic structural characteristics, and environmental
complexity - measured by the number of hospitals within the local area - increases and 2)
there were significant interaction effects between task complexity and structural features.
Assuming that a hospital adds additional medical specialties, the likelihood of adopting
v
an EHR system of the hospital increases under the decentralized decision-making system,
but decreases under the centralized decision-making system. The likelihood decreases
under a high level of IT infrastructure, but increases under a lower level of IT
infrastructure. For the hospitals’ EHR use, there was not any relationship between EHR
use and proposed hospital’s internal features. Thus, alternative measures of EHR use and
internal features were suggested. For EHR interoperability, this study found that 1) the
likelihood of having EHR interoperability increases as task complexity and organic
managerial features increases, and 2) two interaction effects were reported. Assuming
that a hospital adds additional medical specialties, the likelihood of having EHR
interoperability of the hospital increases at a high level of IT staff specialization, but
decreases at a lower level of IT staff specialization. At a high level of environmental
complexity with more than average number of hospitals within the local area, the
likelihood of having EHR interoperability of the hospitals located in the area increases as
IT staff specialization increases. However, the likelihood decreases as IT staff
specialization increases at a lower level of environmental complexity with less than
average number of hospitals within the local area.
In conclusion, this study verified that hospitals’ task, structure, and environmental
features were critical factors affecting the EHR system adoption and interoperability.
However, these factors did not affect EHR use. Different approaches measuring EHR use
and hospitals’ various internal features were recommended. This study’s results can
provide health informaticians, hospital IT managers, and health politicians with new
information about EHR system adoption, use, and interoperability for their innovative
decision-making.
vi
Table of Contents
Acknowledgements ............................................................................................................ i Dedication ......................................................................................................................... iii Abstract ............................................................................................................................. iv Table of Contents ............................................................................................................. vi Appendices ...................................................................................................................... viii List of Tables .................................................................................................................... ix List of Figures .................................................................................................................... x CHAPTER I INTRODUCTION ..................................................................................... 1 1.1. Recent central phenomena in hospitals ........................................................................ 1 1.2. Significance of the work ............................................................................................ 10 CHAPTER II BACKGROUND AND LITERATURE REVIEW .............................. 12 2.1. Study setting............................................................................................................... 12 2.2. EHR adoption, use, and interoperability .................................................................... 14
2.2.1. EHR adoption...................................................................................................... 15 2.2.2. EHR use .............................................................................................................. 16 2.2.3. EHR interoperability ........................................................................................... 17
2.3. Theoretical background ............................................................................................. 18
2.3.1. Contingency theory ............................................................................................. 18 2.3.2. Contingency theory constructs ............................................................................ 20 2.3.3. Task-Technology fit model ................................................................................. 25
2.4. Factors that influence EHR adoption, use and interoperability ................................. 26
2.4.1. Factors that influence EHR adoption .................................................................. 26 2.4.2. Factors that influence EHR use ........................................................................... 27 2.4.3. Factors that influence EHR interoperability ....................................................... 28 2.4.4. Summary of factors influencing EHR adoption, use, and interoperability ......... 30
2.5. Theory-based factors that influence EHR adoption, use, and interoperability .......... 31 2.6. Conceptual model ...................................................................................................... 37 2.7. Objectives, specific aims, and hypotheses ................................................................. 41 CHAPTER III RESEARCH METHODS ..................................................................... 44 3.1. Research design ......................................................................................................... 44
vii
3.2. Study population ........................................................................................................ 45 3.3. Definition of an EHR system ..................................................................................... 48 3.4. Data collection using survey instrument .................................................................... 48
3.4.1. Measuring EHR adoption, use, and interoperability ........................................... 48 3.4.2. Measuring hospital internal features ................................................................... 53 3.4.3. Pilot survey ......................................................................................................... 56 3.4.4. Survey data collection ......................................................................................... 57
3.5. Additional data collection .......................................................................................... 59 3.6. Statistical data analysis .............................................................................................. 62
3.6.1. Binary outcome variables: EHR adoption and interoperability .......................... 64 3.6.2. Numeric outcome variable: EHR use ................................................................. 66 3.6.3. Handling multicollinearity issue ......................................................................... 69 3.6.4. Procedures for any interaction terms of regression model .................................. 69
3.7. Reliability test on survey instruments ........................................................................ 70 3.8. Correlation analysis among the variables .................................................................. 71 CHAPTER IV RESULTS .............................................................................................. 74 4.1. General characteristics of study hospitals .................................................................. 74 4.2. EHR adoption............................................................................................................. 75
4.2.1. Hospital characteristics by EHR adoption .......................................................... 75 4.2.2. EHR adoption and task complexity .................................................................... 77 4.2.3. EHR adoption and hospital structure .................................................................. 78 4.2.4. EHR adoption and environment.......................................................................... 79 4.2.5. EHR adoption, task, structure, and environment ................................................ 80
4.3. EHR use ..................................................................................................................... 84
4.3.1. Hospital characteristics by EHR use ................................................................... 84 4.3.2. EHR use and task complexity ............................................................................. 85 4.3.3. EHR use and structure ........................................................................................ 86 4.3.4. EHR use and environment .................................................................................. 87 4.3.5. EHR use, task, structure, and environment ......................................................... 87
4.4. Interoperability ........................................................................................................... 89
4.4.1. Hospital characteristics by EHR interoperability ............................................... 89 4.4.2. EHR interoperability and task complexity .......................................................... 90 4.4.3. EHR interoperability and structure ..................................................................... 91 4.4.4. EHR interoperability and environment ............................................................... 92
viii
CHAPTER V DISCUSSION .......................................................................................... 97 5.1. Study methods and related issues .............................................................................. 97
5.1.1. Study design ........................................................................................................ 97 5.1.2. Study population ................................................................................................. 99
5.2. Research results ....................................................................................................... 100
5.2.1. Hospital’s internal features affecting EHR adoption ........................................ 100 5.2.2. Hospital’s internal features affecting EHR use ................................................. 105 5.2.3. Hospital’s internal features affecting EHR interoperability ............................. 108
5.3. Study limitations ...................................................................................................... 110 CHAPTER VI CONCLUSIONS ................................................................................. 112 6.1. Study overview ........................................................................................................ 112 6.2. Significant findings .................................................................................................. 113
6.2.1. Findings in EHR adoption ................................................................................ 113 6.2.2. Findings in EHR use ......................................................................................... 115 6.2.3. Findings in EHR interoperability ...................................................................... 115
6.3. Contributions to the field of health informatics ....................................................... 117 6.4. Future study directions ............................................................................................. 118 * Bibliography ................................................................................................................ 121 Appendices Appendix 1. Hospital Information Technology Survey (English) .................................. 131 Appendix 2. Hospital Information Technology Survey (Korean language) ................... 136 Appendix 3. IRB Letter................................................................................................... 141 Appendix 4. University of Minnesota home page for the survey ................................... 142 Appendix 5. Survey introduction and password section ................................................. 143 Appendix 6. Survey main page ....................................................................................... 144
ix
List of Tables
Table 1. Characteristics of mechanical and organic organizational forms ....................... 24
Table 2. Basic characteristics of hospitals in South Korea (beds) .................................... 46
Table 3. Basic characteristics of hospitals in South Korea (foundation type) .................. 47
Table 4. Basic characteristics of hospital in South Korea (geographic distribution) ........ 47
Table 5. Measuring EHR adoption, use, and interoperability ........................................... 51
Table 6. Hospital basic characteristics included in the survey ......................................... 53
Table 7. Measures of factors affecting EHR adoption, use, and interoperability ............. 55
Table 8. Hospital basic characteristics using other information sources .......................... 60
Table 9. The number of specialties in study hospitals ...................................................... 60
Table 10. Measures of environmental complexity ............................................................ 62
Table 11. Reliability of subjective survey questions ........................................................ 70
Table 12. Correlations between variables of all study hospitals ....................................... 72
Table 13. Correlations between variables of hospitals only having EHR adoption ......... 73
Table 14. Basic characteristics of study hospitals ............................................................ 74
Table 15. Basic characteristics of study hospitals by EHR adoption status ..................... 75
Table 16. Hospital’s internal features by EHR adoption status ........................................ 76
Table 17. Impact of task complexity on EHR adoption .................................................... 78
Table 18. Impact of hospital structure on EHR adoption ................................................. 79
Table 19. Impact of environmental complexity on EHR adoption ................................... 80
Table 20. Impact of a hospital’s internal features on EHR adoption ................................ 81
Table 21. Characteristics of hospitals adopting an EHR system by EHR use .................. 85
Table 22. Impact of task complexity on EHR use ............................................................ 85
Table 23. Impact of hospital structure on EHR use .......................................................... 86
Table 24. Impact of environmental complexity on EHR use ............................................ 87
Table 25. Impact of a hospital’s internal features on EHR use ........................................ 88
Table 26. Characteristics of hospitals adopting an EHR system by interoperability ........ 89
Table 27. Hospital’s internal features by EHR interoperability status ............................. 90
Table 28. Impact of task complexity on EHR interoperability ......................................... 91
Table 29. Impact of hospital structure on EHR interoperability ....................................... 91
Table 30. Impact of environmental complexity on EHR interoperability ........................ 92
Table 31. Impact of a hospital’s internal features on EHR interoperability ..................... 93
Table 32. Comparison of study designs ............................................................................ 97
x
List of Figures
Figure 1. Map of South Korea with province identified .................................................. 12
Figure 2. A conceptual model explaining technology adoption ....................................... 38
Figure 3. A conceptual model explaining HIT adoption & use ........................................ 39
Figure 4. Application for EHR adoption using TTF model .............................................. 40
Figure 5. Conceptual model .............................................................................................. 40
Figure 6. Study design flow chart ..................................................................................... 44
Figure 7. Task complexity, decision-making structure, EHR adoptions 1 ....................... 82
Figure 8. Task complexity, decision-making structure, EHR adoptions 2 ....................... 82
Figure 9. Task complexity, IT infrastructure, EHR adoptions 1 ...................................... 83
Figure 10. Task complexity, IT infrastructure, EHR adoptions 2 .................................... 84
Figure 11. Task complexity, IT staff specialization, EHR interoperability 1 ................... 94
Figure 12. Task complexity, IT staff specialization, EHR interoperability 2 ................... 95
Figure 13. Environmental complexity, IT staff specialization, EHR interoperability1 .... 95
Figure 14. Environmental complexity, IT staff specialization, EHR interoperability2 .... 96
Figure 15. Impacts of hospital structure & environment on EHR adoption ................... 115
Figure 16. Impacts of hospital structure & environment on EHR interoperability ........ 116
1
CHAPTER I INTRODUCTION
1.1. Recent central phenomena in hospitals
Medical care generates an extraordinary amount of data, almost all of which has been in
paper-based medical records. Once patients visit healthcare organizations, a variety of
patient information, such as family histories, clinical histories, lab results, x-rays, and
billing records are collected and recorded. This information is used to treat the patient if
he or she revisits the organization. Traditionally, paper-based medical records have
incorporated vast amounts of patient information and have had a dominant role in
medical care.
Thomas S. Kuhn, in his book The Structure of Scientific Revolution, defined a scientific
paradigm as a belief verified and shared in a specific generation. He stated that a
paradigm gradually shifts to the next paradigm through revolutionary activities when
new scientific discoveries are found.1 For example, many people once believed that our
earth was the center of the universe, but we now know differently.
Already we are beginning to see a paradigm shift in the field of health informatics as
more hospitals computerize diverse patient medical records. One of the major trends
recently evolving in South Korean hospitals is replacing the paper-based medical records
(PMR) with electronic health record (EHR) systems. Many hospitals are adopting EHR
systems for diverse functionalities such as reminders, alerts, scheduling, and decision
supports, and they are increasing their capacity for healthcare information exchange
(hereafter interoperability) between many systems.
2
The EHR system was first introduced into South Korean hospitals in 1995.2 According
to a study conducted by Park (2005) and his colleagues, by 2004, 13.3 percent (4/30) of
teaching hospitals in Korea had fully adopted and 53.3 percent (16/30) had partially
adopted an EHR system. Full and partial adoption of an EHR system in non-teaching,
general hospitals was at 7.6 percent (7/92) and 30.4 percent (28/92) respectively. Park
and his colleagues’ study, which focused on hospitals with at least 100 beds, had a 43.1
percent response rate: 122 hospitals participated in the study out of a general hospital
population of 282 hospitals.3 In a nationwide survey conducted in 2004, the EHR
adoption rates for inpatients and outpatients were 19.6 percent (62/324) and 20.7 percent
(65/324) respectively. This study targeted 1,289 hospitals, of which 324 hospitals
responded.4 According to a study conducted in 2003, 63.7 percent of primary care
clinics were adopting electronic medical record (EMR) systems. The study randomly
selected 2,000 primary care clinics from the roster of primary care clinics put out by the
Korean Medical Association; 525 primary care clinics participated in the survey (26.3%
response rate).5 According to a study on the use of EHR functionalities and EHR
interoperability in Korea, the EHR system performed multiple functions – such as drug
interaction alert (52.9% of the hospitals), drug dose alert (52.4%), clinical pathways
(42.3%), drug allergy alerts (25%), and clinical guidelines (20%) – and interoperability
operations – such as provider-provider connection (54.5%), connection with external
organizations such as pharmacies (44.0%) and sending e-mail (18.8%), and integrated
medical records with other hospitals (4.2%).6 This study surveyed medical record
administrators of medical record departments in general hospitals in 2005 and there was
a 59.8% (55/92) response rate.
3
EHR adoption, use, and interoperability in Korea are expected to continually increase
judging from changes occurring in the Unites States. According to a recent study
systematically reviewing research studies completed between 1994 and 2005 in the U.S.,
hospitals’ EHR adoption rates have been predicted to reach between 20 and 40 percent.7
Another recent study shows that 11 percent of hospitals had fully implemented and 57
percent had partially implemented the EHR system in 2007.8 With respect to use of EHR
systems, Felt-Lisk found that hospitals used various EHR functionalities: “electronic lab
results” (88% of hospitals), “electronic clinical notes” (59%), “electronic images
available throughout hospital” (50%), and “electronic lab orders” (49%).9 This study
conducted a nationwide survey to hospitals about the use of EHR functionality by asking
if clinicians at the hospital used any of a list of electronic health record capabilities.
Gans and his colleagues showed which functions of EHR were used in the medical
group practice level. They found that medical groups used “patient demographics”,
“visit/encounter notes”, and “patient medications/ prescription” (p. 1327). EHR systems
with drug formularies were implemented least often, at around 65 percent of respondent
EHR systems. The highest functionality, at 100 percent, was a function providing patient
demographic information.10 With respect to interoperability of an EHR system,
according to a foundation for e-health initiative which has conducted an annual
nationwide survey for health information exchanges since 2004, the percentage of
transmitting data among healthcare organizations was “laboratory result 34% of the
time”, “outpatient episodes 32% of the time”, and “emergency department episodes 30%
of the time”. Study subjects participating in data exchanges were hospitals (71%),
primary care physicians (74%), community health centers (69%), and specialty care
4
physicians (66%).11 Adler-Milstein and her colleagues investigated the degree of
healthcare information exchanges among healthcare organizations and found that nearly
40 percent of respondents did not get any healthcare information exchanges and 50
percent of the rest of the respondent group had only partial information exchanges in
place in the form of “test results”, “inpatient data”, “medical history”, and “outpatient
data” (p.65). 12
This transition from PMR to EHR systems occurs for several reasons such as increased
managerial efficiency and increased effectiveness of healthcare associated with EHR
systems. A patient’s clinical information such as past and current health status can be
easily accessible to the authorized medical professionals to enable their best decision-
making. Various functionalities in EHR systems make it possible for medical providers
to access updated patient information. According to a study in Korea that evaluated the
effect of EHR systems on nursing documentation by comparing documentation before
and after EHR adoption, EHR systems significantly increased direct patient care time
and decreased the time needed for nurses to document patient data.13 Another study in
Korea compared the volume of data on patient care between PMR and EHR systems by
comparing data before and after implementing the EHR system. There was not any
decrease in the data volume.14 A study recently conducted in Korea shows that the
quality of documentation on “the existence of the record” and “level and agreement of
information” was better under an EHR system than PMR system.15 Pizziferri, et al,
conducted a pre-post motion study after implementing an EHR system. They observed
20 physicians working at five primary care clinics and found that physicians thought
EHR improved quality of care, access to patient information, and communication with
5
patients.16 Wong, et al, investigated the impact of computerized documentation on
nursing and patient care time in intensive care units, and they found that EHR adoption
in intensive care units reduced documentation time by 10.9 percent and increased direct
patient care time by 8.8 percent.17 According to a comprehensive literature review study,
many studies showed that EHR systems were accurate, reliable, and relevant to patient
care. 18, 19 Hospital administrations using IT and replacing the paper medical record
systems with electronic record systems not only simplify their work process, they also
save time and space in healthcare organizations.
There are many obstacles and barriers preventing the adoption of the EHR system, use,
and interoperability. Many physicians are worried about data entry, implementation
costs, security issues, and confidentiality of patient information when they use electronic
medical records systems.20 Interface issues incorporating all current diverse systems,
such as laboratory systems and pharmacy systems, into EHR systems may be another
obstacle to implementing EHR systems.21 As mentioned in an empirical study by Payne
et al, factors such as “leadership”, “application functionality”, “speed”, and “data
availability” may be other technical issues.22 These issues set forth a question: how can
we encourage EHR adoption, use, and interoperability by organizations?
There are many different approaches to eliminating these obstacles and barriers to EHR
system adoption and use. One approach is encouraging EHR adoption and use through
various governmental initiatives that are part of the process moving toward producing
better health outcomes. Recent government initiatives are good examples. In December
2005, the Korean government’s Ministry of Health and Welfare (MOHW) founded “the
Center for Interoperable EHR (CiEHR)” to support EHR technology development and
6
adoption in hospitals.23 The CiEHR is a governmental research institute that supports
the MOHW as they develop core EHR technologies for private and public sectors. 24
Changes occurring in the U.S. are even stronger and more practical than those in Korea.
The U.S. federal government has made an effort to increase IT adoption, use, and
interoperability using stimulus money, especially for “meaningful use” of certified
EHRs. The Center for Medicare and Medicaid Services (CMS) and the Office of the
National Coordinator for Health Information Technology (ONC) are now developing
and proposing a definition of the meaningful use of certified EHR technology. They are
considering that “meaningful use” of certified EHR should improve “health care quality”,
“efficiency”, and “patient safety” and have the functionality areas of “disease
management”, “clinical decision support”, “medication management”, “support for
patient access to their health information”, “transitions in care”, “quality measurement
and research”, and “bi-directional communication with public health agencies”.25 In
February 2009, President Barack Obama signed the American Recovery and
Reinvestment Act (ARRA). The ARRA supports IT and EHR use and interoperability to
produce better health outcomes through using and sharing health care information.26 The
Department of Health and Human Services (DHHS) is implementing the Health
Information Technology for Economic and Clinical Health (HITECH) Act to encourage
IT adoption, use, and interoperability, including EHR systems, as a part of the ARRA,.
The DHHS has allocated approximately $19 billion to support its specific plans, $17
billion of which would be used for physicians and hospitals. Starting in 2011, physicians
and hospitals using “meaningful use” of a certified EHR system would receive part of
the allocated money.26 This initiative could persuade many healthcare organizations and
7
professionals to adopt and use EHR systems, which would result in expected positive
outcomes. However, these governmental initiatives may not be enough to accelerate
EHR adoption, use, and interoperability unless we understand the fundamental
mechanism at the healthcare organizational level better. This is because many cultural or
behavioral barriers may exist in organizations.27
Another complementary approach would be to conduct empirical studies to develop a
better understanding of why organizations adopt and use EHR systems in order to
encourage EHR adoption and use. What differentiates organizations that do move
quickly from those that do not? Users of these results such as hospitals managers, health
informaticians, and politicians can use the study outcomes to eliminate obstacles and to
stimulate factors accelerating EHR system adoption and use. There are still many
unknown factors, and thus the research questions for this study focus on the latter
complementary approach.
There have been a relatively small number of studies that relate hospital factors to EHR
system adoption, use, and interoperability in Korea. This is because South Korea has a
short history of EHR adoption. By applying the American Hospital Association’s (AHA)
EHR definition, Park (2006) surveyed Korean general hospitals. In 2005, The Korean
Hospital Association (KHA), the Korean Society of Medical Informatics (KOSMI), and
the Health Insurance Review & Assessment Service (HIRA) conducted a nationwide
survey regarding hospitals’ IT status. Although the survey comprehensively
incorporated hospitals’ structural factors and EHR adoption, the study only reached the
descriptive analysis level on EHR adoption. There have been several other studies on
EHR or IT adoption in Korea. However, they have mainly focused on factors such as
8
changes in the number of medical record administrators after the computerization of
medical records,28 nurses’ experience after implementing hospital information systems,29
a case study from a university hospital after implementing an EMR system,30 and
satisfaction with the EHR system.31 Therefore, it is necessary to conduct a study
investigating factors affecting EHR adoption, use, and interoperability in healthcare
organizations.
Although there have been several studies on factors and barriers related to EHR adoption,
use, and interoperability conducted outside of South Korea, they have several limitations.
First, most of these studies asked for opinions about factors that are barriers to EHR
adoption, such as financial resources, concerns, or worries about patient confidentiality
issues, staff resistance, fear of computers, training and education issues on new systems,
and maintenance of the system.32
Second, other than studies on EHR adoption, there has been a lack of studies focusing on
how the use of EHRs and their interoperability are related to organizational factors. EHR
use and interoperability are concepts recently introduced, and thus, their theoretical
conceptualization and their relationships with other constructs are less developed.
Third, studies of EHR adoption and use utilizing theoretical concepts at an
organizational level are rare. Many scholars in health informatics and management
information systems (MIS) have studied why people accept technology and what factors
affect an individual’s technology acceptance. Common theories these studies use are
Fishbein and Ajzen’s Theory of Reasoned Action (TRA), Davis’s Technology
Acceptance Model (TAM), and Ajzen’s Theory of Planned Behaviors (TPB).34 However,
those studies focused mainly on individuals. Many hospitals are now implementing
9
various IT systems. Why are some hospitals beginning to install EHR systems while
others are not? A mechanism explaining technology acceptance model at the individual
level may not work at the organizational level because hospital level factors are quite
different from individual factors, and decision-making regarding adoption of an EHR
system is determined at the organizational level. In Korea, decisions such as adopting an
EHR system are indeed made at the organization level by the hospitals and clinics. EHR
adoption, use, and interoperability at the organizational level needs to be studied and
applied using current theoretical perspectives developed from outside the academic field
of health informatics, like the technology acceptance model does at the individual level.
Studies which use a theoretical perspective are useful for predicting expected outcomes.
They can provide us with a more fundamental mechanism for phenomena occurring in
healthcare fields.
The objective of this study is to investigate organizational characteristics of hospitals
that affect EHR system adoption, use, and interoperability. The results may be used to
support various governments’ political initiatives through better understanding of
obstacles and barriers to EHR adoption, use, and interoperability. By eliminating those
fundamental obstacles and barriers identified throughout this study, hospitals and
governments can accelerate EHR adoption, use, and interoperability, which could bring
greater managerial efficiency and healthcare effectiveness. This in turn should result in
better health outcomes.
10
1.2. Significance of the work
First, this work has crucial implications for hospital management because it can tell us
which factors are related to EHR adoption, use, and interoperability and what the
precursor conditions are. None of the previous studies in Korea investigated the
relationship between these three constructs and organizational factors. The results may
act as a catalyst for decision making concerning acquiring and using EHR systems in
Korea.
Second, it may enhance healthcare policymakers’ understanding of hospitals’ EHR
adoption and use behaviors. Identifying related factors is important because actual
benefits from the system would occur through hospitals’ adoption of the system, its use,
and healthcare information exchanges with other organizations. Using information
derived from this study, they may design more effective and efficient strategies for
achieving their IT diffusion policy goals.
Third, this study can contribute to theory building on IT adoption, use, and
interoperability at the organizational level. A significant number of studies have focused
on individual adoption behaviors, but relatively few studies have tried to investigate
organizational adoption in health informatics based on theoretical models from other
fields.
Fourth, this study will provide important information on EHR-related issues that can be
used for international comparisons. There have been relatively few opportunities to see
fellow nations’ statuses regarding EHR adoption, use, and interoperability issues.
Through understanding how hospitals responded to EHR adoption in other countries, we
11
can acquire some knowledge about hospitals’ behaviors. This kind of knowledge can be
used for IT dispersion in the healthcare industry.
This study expects that its results will provide hospital managers, scholars, and
politicians working in health informatics fields with useful information on factors
affecting EHR adoption, use, and interoperability in hospitals. This study also expects
that its results can be used to eliminate some of the barriers to adopting EHR systems
and to accelerate EHR system adoption and use, which will indirectly contribute to
improvement of health outcomes in healthcare organizations.
12
CHAPTER II BACKGROUND AND LITERATURE REVIEW
2.1. Study setting
The study setting is South Korea (hereafter Korea). Korea is a country located on the
southern half of the Korean Peninsula in East Asia. It is bordered by North Korea to the
north and neighbored by China to the west and Japan to the southeast.34 Its capital is
Seoul. Korea is 38,023 square miles,35 which is smaller than half the size of the state of
Minnesota (86,943 square miles) 36 in the United States. Its population was 48.6 million
in 2008,37 which is almost 10 times that of Minnesota’s population (5.2 million)38 during
the same year. Korea consists of 16 states: 1 Special District City (Seoul), 6
Metropolitan Cities, and 9 Provinces (Figure 1).
Figure 1. Map of South Korea with province identified 39
13
The Korean government introduced an initial health insurance system in 1977, and
implemented a compulsory health insurance program for the entire population in 1989.
The financial resources of the insurance system consist of contributions paid by the
insured, their employers, and government subsidies. There were three kinds of health
insurance schemes before 2000. One was for ordinary employees, another for
government and private school employees, and the third was for self-employed persons
and rural area residents. However, these three types of insurance schemes were
combined into one insurance system in 2000 and have since been managed by the
National Health Insurance Corporation.
Despite the government’s involvement in controlling the insurance system there was no
law or regulation on electronic medical documents or records in Korea before 2003. The
Korean government revised the Medical Act on March 30, 2003 to say that healthcare
organizations can create and keep electronic medical documents40 (in addition to the
paper medical records and documents) that have a written signature from the healthcare
provider that is applied under the guidelines of the Digital Signature Act. The Korean
Medical Law currently notes that “medical professions or founders of healthcare
organizations can create and keep electronic documents called ‘Electronic Medical
Records’ having a written signature applied by [under the guideline of the] Digital
Signature Act although there is the rule of section 22.” 41 The rule of section 22 requires
healthcare providers to keep various medical patient care documents with their
signatures.
This law is now applied to all hospitals. There are several types of hospitals in Korea.
Korean Medical Law divides hospitals into three categories: 1) general hospitals,
14
hospitals with 100 or more beds and at least eight clinical departments (e.g. internal
medicine, general surgery, pediatrics, obstetrics and gynecology, imaging medicine,
emergency medicine, and clinical pathology), 2) hospitals (hereafter, this study will use
the term, “small hospital” to represent “hospital” under the definition of the Korean
Medical Law, in order to avoid terminological confusion with a general term “hospital”),
medical facilities accommodating 30 or more beds, (this includes dental hospitals and
hospitals of oriental medicine), and 3) long-term care hospitals, facilities which are
under the control of the Social Welfare Law and which provide long-term care.42
Approximately twenty percent of these hospitals have now adopted EHR systems.
However, there has been a paucity of studies about how a hospital’s structural
characteristics affect that hospital’s EHR system adoption, use, and interoperability.
Information about the past status of hospitals that have adopted EHR systems can
support the decision to adopt the system of hospitals that have not adopted an EHR
system. The following sections will describe how current empirical studies define and
measure EHR adoption, use, and interoperability. Then theories that can help explain
why organizations make decisions to adopt and use EHRs will be reviewed. Finally
specific factors that are derived from these theories and that are demonstrated to have an
impact on such decisions will be reviewed.
2.2. EHR adoption, use, and interoperability
This section briefly reviews how current empirical studies define and measure the
concepts of EHR adoption, use, and interoperability at the hospital level.
15
2.2.1. EHR adoption
The AHA performs an annual survey of hospital information technology adoption,
including EHR adoption. It measures EHR adoption at one of three levels: full
implementation, partial implementation, or no EHR system implementation.8,43 The
survey provides a definition of an electronic health record system, but does not provide a
formal definition of adoption.
DesRoches and her colleagues surveyed the adoption of the EHR system in primary care
clinics. They categorized two types of EHR systems: a basic EHR system and a fully
functional system. They measured EHR adoption by looking at whether primary care
clinics acquired and implemented an EHR system, but they also did not define the
concept of EHR adoption.44 Although they measured EHR adoption only in primary care
clinics, but not hospitals, they set forth an example of how current studies measure EHR
systems in healthcare settings.
Jha and his colleagues conducted a nationwide survey of hospitals in which they
investigated the adoption rate of EHR systems in hospitals in 2008. They also did not
formally define adoption, but they categorized it as: basic or comprehensive EHR
system adoption, adopting an EHR system, and no adoption. They defined a
comprehensive EHR system as a system in which all of the hospital’s clinical
department units have implemented an EHR system with various clinical functionalities.
A basic EHR system is one in which only part of the hospital’s clinical units have
implemented an EHR system. If any part of the hospital units had begun to implement
an EHR system, they considered those cases as adopting an EHR system. If they did not
have any EHR system, then they labeled the hospital as not adopting.45 These studies
16
define the concept of EHR adoption in a hospital as any actual acquisition and
implementation of an EHR system by any department in the hospital.
2.2.2. EHR use
The above-mentioned study conducted by DesRoches and her colleagues also focused
on EHR system use measured as self-reported utilization of system functions. According
to their study, 4% of physicians were using a fully functional EHR system of which 97%
of respondents reported that they are using all the functions in the system at least some
of the time. Thirteen percent of physicians were using a basic EHR system and more
than 99% of those physicians reported that they are using all the functions at least some
of the time.44
Most current studies have defined “EHR use” as the existence of implemented functions
in accordance to the Medical Group Management Association (MGMA).46 For an
example, Jha and his colleagues measured use of an EHR system as whether or not it
had specific functionalities such as clinical guidelines, clinical reminders, drug-allergy
alerts, and drug-allergy interaction alerts.45 In 2003, IOM suggested that an EHR system
can be used primarily for patient care delivery, patient care support processes, other
administrative procedures, and secondarily for education and clinical research. They also
set forth that an EHR system should have core functionalities such as decision-support,
results management, electronic communication and connectivity.47 Recently, the CMS
and the ONC defined the meaningful use of an EHR system as the existence and/or
reported use of decision-support, clinical pathways, and interoperability.48 These studies
17
have not presented clear definitions but rather have relied on self-reports of the functions
of an EHR system that are in place and being used by hospital personnel.
2.2.3. EHR interoperability
The Healthcare Information and Management Systems Society (HIMSS) defines
interoperability as “the ability of health information systems to work together within and
across organizational boundaries in order to advance the effective delivery of healthcare
for individuals and communities.” 49 The U.S. Government Accountability Office
defines interoperability as “the ability of two or more systems or components to
exchange information and to use the information that has been exchanged (p.1).” 50
Bouhaddou and his colleagues also define interoperability as “the ability of two or more
health care information systems to exchange information and to use the information that
has been exchanged (p.174).”51 These definitions clearly illustrate that EHR
interoperability is the ability of an EHR system to exchange patient information with
other systems.
The National Library of Medicine (NLM) and Health Level 7 (HL7) developed a survey
that measured interoperability by asking “what type(s) of information are being
exchanged?” The respondent answered the question with either a yes or no on each
clinical item, such as family or patient history, patient condition summary, discharge
summary, and other lab and procedures. 52 Using such a measure, the e-Health Initiative
characterized interoperability of EHRs in 2007 as exchange of laboratory result (34%),
outpatient episodes (32%), and emergency care episodes (30%).11 This study could not
18
find any other previous studies operationally defining and measuring interoperability in
hospitals.
2.3. Theoretical background
There are many organizational theories which can provide insight into why
organizations make decisions to implement and use EHRs. This study selected two of
these theories because they are particularly applicable to our research question. This
section will introduce those theories, their critical concepts, and arguments.
2.3.1. Contingency theory
The basic ideas of contingency theory were introduced by Lawrence and Lorsch.55 They
found that environmental uncertainty was associated with structural differentiation such
as departmentalization and the division of labor. This structural differentiation was also
inversely related to integration. Organizations having high integration efforts achieved
high performance as measured by profit changes, sales volumes, and manager’s
subjective appraisals.53 Based on these findings, they argued that organizations would
be effective in achieving their performance goals when their internal features, such as
subunits, and their processes, such as integration efforts, are consistent with external
environments.53 Jay Galbraith (1973) stated that “there is no one best way to organize”,
but that “any way of organizing is not equally effective.”54 This means that
organizational performance depends on various structural factors including environment.
Scott (2003) explains this theory with this hypothesis: “organizations whose internal
19
features best fit the demand of their environments will achieve the best adoption”
(p.96).55
However, some scholars use a slightly different terminology, strategic contingency
theory, when intra-organizational changes occur in response to the environment. For
example, Hickson and his colleagues note that “the theory relates the power of a subunit
to its coping with uncertainty, substitutability, and centrality, through the control of
strategic contingencies for other dependent activities, the control resulting from a
combination of these variables” (p.216).56 D. Knoke also illustrates strategic
contingency theory with intra-organizational or subunit changes caused by external
environment. Organizational environments affect organizations having many subunits.
Subunits that prove more successful in dealing with the environment will achieve greater
dominance over other subunits and the organization’s resources and will have a greater
effect on organizational rules and procedures.57 For example, if legal claims from
consumers occur frequently in an organization, the legal subunit manager can take a
dominant position and critically affect other subunits that are producing and maintaining
products. As an example of contingency theory, Orlikowski and Barley state that “the
more complex and unpredictable the technology, the more likely were organizations to
adopt an organic rather than a mechanistic structure” (p.148).58 What does this theory
tell us about adoption of an EHR system? This study argues that there are certain
hospital internal features that promote and correspond well with EHR adoption, use, and
interoperability. Hospitals with those characteristics are more likely to adopt an EHR
system and use the system more than the other hospitals without those characteristics.
The arguments regarding these predictions will be described in the next section in detail.
20
2.3.2. Contingency theory constructs
There are three theoretical concepts frequently dealt with in contingency theory: task,
structure, and environment. These concepts can be analyzed into several further
theoretical constructs. For example, task can be broken down into task complexity,
uncertainty, and interdependency. Among the several task constructs, only task
complexity is applicable to our research question because EHR adoption, use, and
interoperability may be critically affected by task complexity, as illustrated by Scott
(2003) and Flood (1987).
With respect to hospital structure and environmental factors, four constructs from
contingency theory focusing on IT departments in hospitals are applicable:
decentralization of decision-making, IT staff specialization, IT infrastructure, and
organic form (versus mechanical form). This is because these four constructs are
frequently mentioned as important factors related to affecting an organization’s
technology use.55,59,60 Additionally, this study also included one environmental factor,
environmental complexity, because the constructs above do not cover environmental
factors.
Task complexity: Task or technology includes the hardware to make products and the
knowledge of employees in the organization (Scott, 2003). Harvey investigated the
relationship between organizational task and organizational structure using 43 industrial
organizations and found that organizations producing specific products had a higher
number of subunits, levels of authority, and ratios of managers and supervisors to total
21
personal than the others.61 These studies tell us about that the various task characteristics
that are a major determinant of organizational structure, which is described in the
following sections.
Decentralization of decision-making system: Decentralization of decision-making
generally means that the authority to make decisions is dispersed throughout many
employees.62 By giving managerial decision-making power to appropriate managers,
organizations can achieve managerial efficacy and organizational effectiveness.
Decentralization of decision-making can be a major determinant of organizational
technology adoption. Hage and his colleague defined decentralization as the extent to
which employees can participate in decision-making about important organizational
programs and policies.63 In a study, they measured the degree of participation of
department heads in an organization’s staff hiring, promotion, program, and policy, and
found that the degree of decentralization of decision-making was positively associated
with an organization’s innovation. They measured innovation as the rate of new program
adoption for a certain period of time.63 In another study, they studied work processes and
organization structure using organizations related to healthcare and interviewed
supervisory personnel at managerial levels. They found that organizations with routine
work-flow had the greater centralization of decision-making system.64 These study
results indirectly suggest that hospitals with decentralized decision-making systems may
be more likely to have complex workflow processes and environments in which
innovation can frequently occur.
22
Specialization: Daft defined specialization as the extent to which organizational tasks
are divided into subunits according to their specialties and responsibilities.65 Scott
illustrated specialization as the degree to which positions in an organization can be
grouped into a certain unit based on similarity of goals, work processes, consumer types,
locations, etc.55 Thus, specialization is related to organizational departmentalization or
subunit as well as the organizational structure. Specialization is sometimes called the
“division of labor” in organization studies.65
There have been many studies on the relationship between specialization and
organizational structure. Main study trends focus on the relationship between
specialization and span of control and between specialization and organizational
integration or coordination. This is because if specialization increases, then an
organization is subdivided into subunits. While narrowing down tasks into subunits
increases work efficiency by giving employees a narrow span of control through limiting
their cognitive capacities, the organizations are more likely to experience difficulty of
control and to confront conflict issues due to increased departmentalization. 66 They thus
need more coordination and integration.67 Consequently, in contingency perspective,
organization efficiency is high when an organization emphasizes coordination or
integration under high specialization. If hospitals are more specialized, then they may be
more likely to adopt an EHR system for integration because EHR systems have various
coordination functions.
Infrastructure: In the studies of the contingency theory, the construct of infrastructure
have not been well developed and studied. Fombrun only defined “infrastructure” as
interconnected core structure that an organization has to deal with and engage in to
23
maintain its works over time.68 He included all feasible sets of resources related to an
organizations’ production into the boundary of infrastructure, including technology and
the various schemes of task. Due to the lack of an empirical study in this area, this study
cannot predict a trend. However, in the contingency perspective, high IT infrastructure
may promote an EHR system adoption because if hospitals have a high degree of IT
infrastructure, then their past experience acquired from previous IT installation may help
them to more easily adopt an EHR system.
Organic or mechanical structure: Burns and Stalker (1961) devised the concept of
mechanical or organic structural forms through the observation of twenty industrial
firms.69 They conceptualized two types of organizational structures: organic or
mechanical. Burns and Stalker observed that the mechanical organizational structure
achieves the best performance under a stable environment. The organic structure is
preferred when there is a high level of environmental change.69 According to their
conceptualization, organic organizations tend to have the following characteristics:
decentralization of decision-making, flexible rules and procedures, frequent mutual
reliance among employees, frequent upward or lateral communications, voluntary
participation of extra activities among employees, and so on. In contrast, organizations
are considered mechanical if their organizational structures do not have these
characteristics. These characteristics are briefly presented in Table 1.
This study asserts that hospitals with a more organic management structure may be more
likely to adopt an EHR system because implementation of new technology, such as EHR
systems, requires acceptance of diverse ideas or needs from employees internally and
externally through lateral communications, which corresponds to the general
24
characteristics of organic management structure. Further specific predictions related to
this construct will be argued in the following section.
Table 1. Characteristics of mechanical and organic organizational forms Items Mechanical form Organic form
CommunicationHierarchical emphasis – predominantly vertical communication
Horizontal emphasis – high level of lateral communication within organizational networks
Decision Centralized decision-making Decentralized decision-making
Roles Well-defined specialized roles Loosely defined and less specialized roles
Rules High reliance on standardized rules and procedures
High reliance on mutual adjustment between co-workers
Source: J. Child (2005), p.26.
Environmental complexity: Tung (1979) defined environmental complexity as “the
number and heterogeneity/diversity of factors and components that the focal unit has to
contend with in decision making” (p.675). She analyzed 64 departmental units of 21
companies and found that departments become more organic in structure as
environmental complexity increases.70 Clark and his colleague defined environmental
complexity as the “degree to which factors in the organization’s environment are few or
great in number and similar or different from one another.”(p.29)71 Duncan set forth the
components of external environment increasing complexity as customers, suppliers,
competitors, and political issues.72 Using these components, he conceptualized
environmental complexity as the simple-complex dimension based on the number of
components and static-dynamic dimensions, changing characteristics of these
components over time. Thus, environmental complexity generally means the degree to
which an organization has to deal with various internal and external factors and
25
resources surrounding it. Previous studies have shown that environmental complexity
affects organizational structure, but they do not describe how environmental complexity
affects technology adoption such as EHR adoption. This prediction will be argued and
described in a following section.
2.3.3. Task-Technology fit model
The Task-Technology Fit (TTF) model was proposed and developed by Goodhue and
Thompson. 73 The TTF model assumes that individuals perform best when their task fits
well with available technology.74 Goodhue and his colleague analyzed users of 25
different information technology systems from two companies and investigated whether
TTF status affects organizational performance. They found that TTF status impacted the
quality of data, timeliness, and the IT system’s relationship with users and was
positively associated with organizational performance measured by perceived
effectiveness and user productivity. 75 Ammenwerth and her colleagues investigated TTF
status among individual users, their tasks, and technologies in hospitals. They observed a
nursing documentation system used in dermatologic, pediatric, and psychiatric wards.
They found that immediately after adopting the nursing documentation system, TTF
status in the dermatologic and psychiatric wards was better than that of the pediatric
ward. 76 These studies indirectly suggested that hospitals with high task complexity
would adopt and use EHR systems because high task complexity fits well with
information processing capabilities of EHR systems. High task complexity requires
hospitals to use coordinated care due to the task complexity. An EHR system can help
them to provide such coordinated care because it has various coordination functions.77 If
26
hospitals with lower task complexity invest in an EHR, it will be an over-investment and
result in managerial inefficiency.
2.4. Factors that influence EHR adoption, use and interoperability
In this section, this study will look at the previous studies related to EHR adoption, use,
and interoperability. It provides the opportunity to see other factors affecting EHR
adoption, use, and interoperability.
2.4.1. Factors that influence EHR adoption
Schoenman conducted a national survey of rural hospitals (sample size 238) and found
that stand-alone, small size, and critical access hospitals were less likely to adopt an
EMR system than other hospitals.78 Although the study was a nationwide survey, the
study was limited by the fact that it only looked at rural hospitals. McCullough analyzed
data from various sources including the HIMSS’s Dorenfest dataset and found that the
scope of services was not related to information system adoption (p.96, p.157).79 The
AHA conducts an annual survey of hospitals’ information technology. According to a
report conducted in 2007, the main factors affecting EHR adoption were the size of
hospital, location of hospital (urban or rural), and teaching status.8 Jha and his
colleagues investigated EHR system adoption in U.S. hospitals through a comprehensive
nationwide survey and also illustrated factors affecting the EHR system adoption in
which hospital size, teaching status, affiliation status, urban location, and existence of
coronary care units were related to EHR adoption. One of the most important barriers to
27
EHR system adoption was the financial issue of purchasing an EHR system.45 These
empirical studies suggest that various structural factors such as size, affiliation status,
hospital location, and teaching status are possible predictors of EHR adoption.
Simon and colleagues also investigated organizational factors affecting EHR adoption in
office practices. They found that practices affiliated with hospitals and practices
teaching medical students are more likely to adopt EHR systems. 80 Healthcare Financial
Management Association surveyed senior healthcare finance executives at hospitals to
investigate barriers to EHR adoption. They received the following responses: “lack of
national information standards and code sets” (62%), “lack of available funding” (59%),
“concern about physician usage” (51%), and “lack of interoperability (50%)” (p. 2).81
This study suggests that the relevance of perceived barriers to adopting an EHR system
may be another critical factor influencing EHR adoption in hospitals.
2.4.2. Factors that influence EHR use
Miller and his colleagues surveyed physicians and EMR managers in 30 physician
organizations to investigate barriers affecting the use of EMR systems. They did not
provide a formal definition of the use of EMR systems, but measured it by whether an
EMR system had the following functions: record viewing capability, documentation and
care management, ordering, messaging, analysis and reporting, patient-directed
functionality, and billing. They reported that patient-directed functions such as e-mail
messaging, reminders, and scheduling were very limited compared to the use of the
other functions. The main barriers to EMR use were mostly financial issues, such as
28
investment costs, low rate of return, and time costs, and additional technological issues,
such as the difficulties of technology management and support and the negative attitude
of physicians (p.119).82
Jha and his colleagues also studied the use of EHR systems in hospitals. They measured
the use of EHR systems by the presence or absence of various EHR functions. There
was a high variation in functions reported. Many EHR systems provided access to test
and imaging results, but did not report using decision-support functions such as clinical
guidelines, clinical reminders, or drug alerts. They reported EHR system adoption and
hospital characteristics, but did not report how they were related to the use of EHR
systems.45 These study results indicate that there has been a relative paucity of studies
investigating how organization factors affect EHR system use. This might be due to the
difficulty of defining the concept of EHR system use. Some studies have investigated
barriers to EHR system use, but findings are similar to the ones in which factors affect
EHR system adoption.
2.4.3. Factors that influence EHR interoperability
Disclosure of patient and provider information, financial burden, additional work load,
and training issues for interoperability may be some critical factors affecting EHR
interoperability. The CMS and the Substance Abuse and Mental Health Services
Administration had a conference on January 24-25, 2007, in which they discussed
interoperability issues in health information systems for mental health, Medicaid, and
treatment for substance abuse. They concluded that major barriers to interoperability
were fear of provider performance and patient privacy disclosure, financing restraints,
29
and additional work.83 The Health Sciences Center at the University of Colorado
investigated the interoperable status of EHR systems in post-acute care and long-term
care facilities and found that physicians’ usage and costs related to interoperable EHR
development and training issues due to high turnover rates were considered important
factors affecting interoperable EHR adoption and development.84
Collaborations and coordination among the users, department managers, and related
people may be other critical factors affecting EHR interoperability. D. Brailer reported
lessons learned from the failure of the Santa Barbara County Care Data Exchange
project that started in 1999 and ended in 2006. He emphasized the importance of an
organic or bottom-up approach to the HIE project with a human factors focus reflecting
end-user’s needs, considerations of workflow and stakeholder diversity.85 Miller and his
colleague described two successful health information exchange projects. According to
their report, two regional health information organizations (RHIOs), the Indiana
Network for Patient Care and the Northwest (Spokane, WA) RHIOs, have had
successful HIE projects. Both RHIOs allow local healthcare providers to access patient
information such as lab tests and medications and use them for patient care. One
important factor to the success of the HIE project reported by authors was voluntary
collaboration of participating organizations and provider networks.86 Although
interoperability is a characteristic of an EHR and HIE is the process of exchanging
information between systems which is not necessarily electronic, but hopefully
electronically interoperable, there is a common characteristic in which both need
members’ collaboration and cooperation for success. These studies suggest that
stakeholders’ participation, coordination among participants, and organic structure of
30
hospitals may be critical factors affecting EHR system interoperability. Without
collaboration and coordination of diverse professional groups in hospitals, hospitals may
not be willing to develop interoperable EHR systems. Collaboration efforts from local
provider networks may be as important for EHR interoperability as cooperation of
professional groups in hospitals.
These empirical studies also suggest that there might be some common characteristics
affecting EHR adoption and EHR interoperability. For example, hospitals may hesitate
to seek health information exchange due to a lack of funding or concerns about
disclosure of individuals’ information. These barriers were also found in a study on
factors affecting IT adoption such as EHR systems.87,88 Although the studies provide
information on barriers to EHR interoperability, the information is often the opinion of
experts and seldom the results from empirical studies.
2.4.4. Summary of factors influencing EHR adoption, use, and interoperability
First, EHR systems or IT adoption was negatively associated with stand-alone, small
size, critical access hospitals, but positively associated with teaching status of hospitals,
high IT infrastructure, large size, urban location, and high case mix index (complexity of
patients treated). Second, EHR systems or IT adoption was not significantly associated
with high-tech services of hospitals, scope of services, types of services. Third, EHR
systems or IT adoption may be negatively associated with hospitals having the following
characteristics: lack of available funding, and concern about physician usage on the
system. Fourth, EHR adoption, use, and interoperability may be positively associated
31
with a hospital’s internal collaboration of employees and community network’s
cooperation.
2.5. Theory-based factors that influence EHR adoption, use, and interoperability
Based on findings from the literature and both contingency theory and Task-technology
Fit Model, this study argues that EHR adoption, use, and interoperability is a function of
a hospital’s task complexity, decentralization of decision-making, IT staff specialization,
IT infrastructure, organic structural forms, environmental complexity, and their
interactions.
Task complexity: In task-technology fit perspective, an EHR system fits well in
hospitals with high task complexity. Hospitals having high task complexity will require
high information processing demands, which will lead the hospitals to invest more in
information technologies including EHR systems because investment in an EHR system
increases the hospital’s capacity to deal with more complex information processing. In
contrast, hospitals with a lower task complexity would not have as great a need to invest
in an EHR system because the lower information processing demands associated with
lower task complexity would make it an unnecessary investment. If they invest some
financial resources into an EHR system, this would result in inefficient resource use. For
this reason, they would not invest in an EHR system. A hospital’s task complexity
would be one of the critical factors of EHR adoption, use, and interoperability. Wang
and his colleagues analyzed data from 1,441 hospitals and investigated factors affecting
adoption of clinical information systems. They found that a case mix index (a measure
of complexity of patients treated) was related to the adoption of the clinical information
32
system.89 The case mix index can be considered as a proxy variable for task complexity
of this study. Their study result provides us a meaningful indication that task complexity
measured by the diversity of the medical specialties may also be related to EHR
adoption, use, and interoperability.
Decentralization of decision-making: In contingency theory perspective, an EHR
system fits well in hospitals with decentralization of decision-making system in terms of
adoption, use, and interoperability. Suppose there are hospitals that allow departmental
managers to actively participate in the hospitals’ major decision-making. These hospitals
would have a more decentralized decision-making system. Hospitals where this is not
allowed have what is called a centralized decision-making system. The possibility of
delivery of new ideas and suggestions including installation of EHR systems to a high
level manager or chief executive officer would more frequently occur in a higher
decentralized system because the lower level managers have more influence. We can see
evidence of this in studies in which IT innovation more frequently occurs in
decentralized decision-making systems. Damanpour conducted a meta-analysis
investigating the relationship between organizational factors and innovation and found
that centralization of decision-making was negatively related to organizational
innovation.134 Accordingly, it is possible that hospitals with decentralized decision-
making systems would be more likely to adopt and use an EHR system than the other
hospitals.
IT staff specialization: Moch and his colleagues investigated factors facilitating the
adoption of innovation through analyzing data from 489 hospitals. The two dependent
variables were the adoption of a new medical technology system for diagnosis and
33
treatment and the adoption of an electronic data processing system for hospital
management. One of the major independent variables was specialization measured by
the number of medical specialty areas. They found that specialization was significantly
associated with information system adoption.90 Thus, this study proposes that hospitals
with specialization would be more likely to adopt and use EHR systems.
IT infrastructure: In hospitals with high IT infrastructure, their employees certainly
would have more experience from previous and current IT implementations. This can
accelerate a hospital to adopt an EHR system compared to hospitals without IT
infrastructure and experience. Empirically, Chau and his colleagues investigated factors
affecting the adoption of IT innovation and found that there was a positive, but not
significant relationship between current IT infrastructure and the adoption of IT
innovation.91 However, Anderson and her colleagues found that current IT infrastructure
was significantly associated with EMR adoption in primary care clinics.92 Although the
unit of analysis was primary care clinics, their study indirectly suggests that IT
infrastructure may also influence adoption of an EHR system in hospitals just as it does
for EMR systems in primary care clinics.
Organic or mechanical forms: A hospital is an organization composed of diverse
specialties and different departments. Installation and use of EHR systems inevitably
require cooperation among various organizational subunits. They also need to coordinate
with diverse subunits through frequent lateral communication. If the installation of an
EHR system is mandated by a top manager without consideration of users’ needs or
opinions, then there would be a high possibility of failure. In contrast, hospitals with an
organic network structure characterized by horizontal communication or network-based
34
installation of EHR systems have a higher possibility of reflecting user-level needs
because frequent horizontal communication can reflect their needs into the EHR system.
Aiken and Hage analyzed a panel data from 16 health and welfare organizations and
investigated the relationship between organic structural characteristics and
organizational innovation. They measured organic features by the diversity of
occupational structure, extra activity of staff members, frequency of meetings, upward
communications, inter and intra-departmental communications and innovation by the
number of new programs and services previously successfully implemented. They found
that organic structural features were positively related to innovation.93 Meadows
investigated organic features of small group structure based on the Burns and Stalker’s
concept and innovativeness, which was measured by group members’ perception on
innovativeness of their work process and management and found a similar result.94 Hull
and his colleague analyzed data from 110 factories and found that innovation, measured
by the number of patent applications adjusting for the number of employees, was high
when there was a more organic rather than a mechanical socio-technical structure. They
measured organic features by hierarchical levels, centralization of decision-making,
degree of control, and no flexibility of waiting-time in assembly line, which was
negatively associated with organizational innovation.95 Thus, adoption and use of an
EHR system would be more likely to occur in an organically structured hospital rather
than a mechanically structured hospital.
Environmental complexity: In a contingency theory perspective, EHRs fit well in
hospitals with high environmental complexity for the following reason. Under a highly
competitive environment, hospitals are more likely to invest in IT because of high
35
competition. If they did not invest in IT, they would fall behind other hospitals and other
competition. Thus, hospitals located in highly complex environments with many
competitors may be more likely to adopt and use an EHR system simply to keep up with
the competition.
Burke, et al investigated the relationship between a hospital’s IT adoption and market
competition. The study used the 1999 Dorenfest database and the 1998 American
Hospital Annual Survey and found that a hospital’s market competition measured by the
Herfindahl Hirschman Index (HHI) is related to IT adoption measured by administrative,
clinical, and strategic IT.96 Baker, et. al., investigated the relationship between Health
Maintenance Organization (HMO) market share and technology adoption measured by
MRI availability per 1,000 populations and found that HMO market share was
negatively related to technology adoption.97 Friedman and his colleagues, through
surveying informants such as the chief executive officer and chief financial officer in
hospitals, studied the relationship between technological and environmental factors with
respect to adopting new medical technology and MRI acquisition. The study looked at
94 hospitals, 52 of which were in California and 42 of which were in either Oregon or
Washington. They found that under a turbulent environment (California), hospitals with
early MRI adoption were significantly affected by technological competition, measured
through hospitals adopting similar technologies. However, under a stable environment
(Oregon/Washington), technologic competition did not play an important role in MRI
adoption. 98
These studies indirectly suggest a possible explanation of the relationship between
technology adoption and market competition. Adopting information technologies may
36
increase a hospital’s competitive value because there is the possibility of better quality
of care and greater managerial efficiency. This gives a good impression to consumers
who often believe that modern technologies produce a higher quality of care. Hospitals
located in less competitive environments thus have less motivation to invest in IT,
including EHR systems, compared to hospitals located in competitive environments.
Therefore, environmental complexity measured by market competition may be
positively related to EHR adoption, use, and interoperability.
Task complexity/structure interaction: Scott reports that organizations with high task
complexity are more likely to have structural specialization and decentralization (Scott,
2003). This is because delegating some authority and responsibility through
specialization and decentralization would be better for performance when an
organization’s task complexity is high. When organizations have specialization and
decentralization, organizational innovation is more likely to occur. From an
organizational perspective, introducing EHR systems can be seen as an innovative action,
and thus, specialized and decentralized hospitals may be more likely to adopt an EHR
system. In contrast, highly centralized hospitals may be less likely to adopt an EHR
system. Brynjolfsson and Hitt (2000), through conducting a survey of approximately 400
firms, found that the degree of IT use had a positive relationship with the degree of
authority delegation to individual employees or teams.99 Thus it is expected that task
complexity may affect EHR adoption, use, interoperability through various structural
characteristics of hospitals proposed in this study.
Environmental complexity/structure interaction: Pfeffer and his colleagues, analyzed
data from 38 small manufacturing organizations in order to investigate the relationship
37
among technology, organization structure, and environmental competition. They found
that environmental competitiveness affects production technology, such as the number
of products, designs, and production processes. Moreover, the relationship between
organizational structure and production technology was dependent on environmental
competitiveness. The number of products increases as departmentalization increases
under competitive environment.100 This study indirectly suggests that there may be some
relationship among technology, structure, and environment. Although this study was not
concerned with EHR adoption, the study suggests that one needs to consider the diverse
relationship between environment and organizational structure with respect to EHR
adoption, use, and interoperability.
Summary of prediction: With respect to the relationship between the three outcome
variables and theoretical constructs proposed in this study, the following relationships
were expected. EHR system adoption, use, and interoperability will be positively related
to task complexity, decentralization of the decision-making system, IT staff
specialization, IT infrastructure, an organic structural management style, and
environmental complexity in hospitals. There will be interaction effects between
technology and structure and between structure and environment with respect to EHR
adoption, use, and interoperability.
2.6. Conceptual model
The following section describes how this study conceptualized each construct from
theories and empirical studies and set forth the final conceptual model for this study. A.
Kazley and her colleague studied organizational and environmental determinants of
38
EMR adoption and developed a conceptual model for her study based on the resource-
dependent theory.101 The resource-dependence theory explains organizational changes
with power-dependence relationships between organizational factors with external
environment. They conceptualized that EMR adoption is a function of organizational
and environmental factors (Figure 2). Although their study focuses on EMR adoption,
the conceptual model can be used for predicting EHR adoption since the concepts are so
similar.
Figure 2. A conceptual model explaining technology adoption
D. Blumenthal and his colleagues also proposed a conceptual model explaining HIT
adoption and use. They argued that HIT adoption and use are critically affected by
technology, organizational factors, legal and regulatory factors, and financial incentives
(Figure 3). This conceptual model is similar to the Kazley model because legal,
regulatory, and financial incentives from governmental regulation settings can be
considered as one of the environmental factors.
These two conceptual models suggest that EHR system adoption and use are a broad
function of organizational structure, technology, and environmental factors. However,
these conceptual models have several limitations. First, these models cannot explain
Environmental factors - Competition - Rurality - Munificence - Uncertainty
EMR adoption Organizational factors - Size - System affiliation - Ownership - Medicare payer mix - Financial resources - Teaching status
Source: Kazley et. al (2007)
39
organizational changes (e.g. adopting EHR systems) occurring due to intra-structural
factors. For example, some hospitals may adopt an EHR system because they need to
deal with high information processing occurring from high task complexity. Task
characteristics can be a critical factor affecting EHR adoption and can replace the
construct of technology as in Figure 3.
Figure 3. A conceptual model explaining HIT adoption & use
Both constructs have indeed been studied with an interchangeable concept in
organization theory (Scott, 2003). To this point, incorporating a TTF model would be
useful to overcome this limitation because the model directly explains the relationship
between task and technology, which is in this case EHR adoption in hospitals (Figure 4).
FIT: adoption or not
Source: Modified from Goodhue et. al (1995)
Technology characteristic Task characteristics
Technology (EHR SYSTEM)
HIT adoption & use
Source: Modified from D. Blumenthal et. al (2006)
Technology
Organizational Legal & regulatory
Financial incentives
Environmental factors
40
Figure 4. Application for EHR adoption using TTF model
Second, these models also did not consider any interaction effects among those
constructs. Several empirical studies have shown that organizational structures are
critically affected by task and environmental factors. For example, Flood and her
colleagues assert that tasks in the work procedure and the environment are major
determinants to hospital structure (Flood, et al., 1987, p.26). Barley (1990), by citing
Woodward’s argument, noted that tasks on production systems are direct determinants
of organizational features such as the formalization of regulations and procedures, span
of control, and centralization of decision-making authority.102
Figure 5. Conceptual model
41
One way to overcome the second limitation is by incorporating contingency theory into
the model because contingency theory assumes some relationship between task and
structure and environment and structure. Incorporating these conceptual models and
limitations mentioned above produces a new conceptual model (Figure 5). Constructs
illustrated in the model were discussed in the literature review and proposed in the
hypotheses.
2.7. Objectives, specific aims, and hypotheses
The objective of this study is to explain Korean hospitals’ EHR adoption, use, and
interoperability by examining the influence of a number of hospital-related factors
including hospital task, structure, and environment derived from both theoretical models
and prior research. This study has the following specific aims and hypotheses.
1) Specific aim 1: To assess the relationship between internal features of hospitals (task,
structure, and environment) and EHR adoption after adjusting for hospital characteristics.
HA1: Hospitals with high task complexity were more likely to adopt an EHR system.
HA2: Hospitals with decentralized decision making, high IT staff specialization, high IT
infrastructure, and an organic managerial structure (versus mechanical structure) were
more likely to adopt an EHR system.
HA3: Hospitals with high environmental complexity were more likely to adopt an EHR
system.
HA4: The relationship between task complexity and EHR adoption would be dependent
on hospital structure.
42
HA5: The relationship between hospital structure and EHR adoption would be
dependent on environmental complexity.
2) Specific aim 2: To assess the relationship between internal features of hospitals (task,
structure, and environment) and EHR use after adjusting for hospital characteristics.
HU1: Hospitals with high task complexity were more likely to use an EHR system.
HU2: Hospitals with decentralized decision making, high IT staff specialization, high IT
infrastructure, and an organic managerial structure were more likely to use an EHR
system.
HU3: Hospitals with high environmental complexity were more likely to use an EHR
system.
HU4: The relationship between task complexity and EHR use would be dependent on
hospital structure.
HU5: The relationship between structure and EHR use would be dependent on
environmental complexity.
3) Specific aim 3: To assess the relationship between internal features of hospitals (task,
structure, and environment) and EHR interoperability (hereafter: interoperability) after
adjusting for hospital characteristics.
HI1: Hospitals with high task complexity were more likely to have interoperability.
HI2: Hospitals with decentralized decision making, high IT staff specialization, high IT
infrastructure, and an organic managerial structure (versus mechanical structure) were
more likely to have interoperability.
43
HI3: Hospitals with high environmental complexity were more likely to have
interoperability.
HI4: The relationship between task complexity and interoperability would be dependent
on hospital structure.
HI5: The relationship between structure and interoperability would be dependent on
environmental complexity.
44
CHAPTER III RESEARCH METHODS
3.1. Research design
This study uses a non-experimental, retrospective cross-sectional study design to
investigate the relationship between internal features of hospitals including environment
and EHR adoption, use, and interoperability (Figure 6).
Figure 6. Study design flow chart
There are three outcome variables: EHR adoption, use, and interoperability. After
categorizing the hospitals into two groups (those which have adopted and those which
Study Hospitals
EHR adoption analysis
EHR use analysis
Group A: EHR adoption Group B: No EHR adoption
EHR USE
No EHR interoperability (Group 2)
EHR interoperability Analysis
Yes No
Yes No
Does your hospital have an EHR SYSTEM?
How many functions does the EHR SYSTEM have?
8 functions
Does the EHR system have INTEROPERABILITY?
Comparison
Comparison
EHR interoperability (Group 1)
Comparison
45
have not adopted an EHR system) and measuring possible causal factors retrospectively,
this study compares those factors between the two groups. In the case of EHR use and
interoperability, this study selects and analyzes only those hospitals which have adopted
an EHR system. The following sections will describe the details of the research design.
3.2. Study population
This study’s unit of analysis is the individual hospital. According to the hospital roster
booklet103 published by the Korean Hospital Association (KHA), there were 1,928
hospitals of all kinds as of December 31, 2007. These included 335 general hospitals,
852 small hospitals, 495 long-term care hospitals, 132 mental hospitals, 57 geriatrics
hospitals, 22 rehabilitation hospitals, 19 veterans’ hospitals, and 6 ophthalmic hospitals.
In addition, there were 10 hospitals that were run directly by the central government for
special purposes: 4 hospitals for pneumoconiosis, 3 hospitals for tuberculosis, and 3
hospitals for leprosy.
Among these hospitals, this study excludes the hospitals directly run by the central or
local governments and military hospitals because the governmental decision-making on
EHR adoption would affect all those hospitals equally. This would result in a high
correlation of those constructs among these hospitals. Second, this study also excludes
long-term care hospitals, mental hospitals, and geriatric hospitals because most of them
are regulated under the Social Welfare Law and the Mental Health Law in Korea, which
has slightly different structural requirements than hospitals under the Medical Law. For
instance, Social Welfare Law and Mental Health Law are more lenient than Medical
Law when it comes to regulating the hiring of medical staff. Third, this study also
46
excludes hospitals based on single specialties such as rehabilitation or ophthalmic
hospitals. They were also structurally different from the majority of general hospitals
and small hospitals because they were providing only specific specialty care services.
Table 2. Basic characteristics of hospitals in South Korea (beds)
Hospital size All hospitals General hospitals Small hospitals
Hospital population 1,063 285 778 Beds Mean 187.8 411.7 105.8 Median 118.0 297.0 87.0 Max 2,200.0 2,200.0 559.0 Min 30.0 100.0 30.0 Std Dev 215.3 302.6 69.0 Bed size 30 – 99 45.0 0.0 61.5 100 - 199 25.6 17.2 28.5 200- 299 15.7 36.0 8.2 300 -399 2.5 6.3 1.2 400 -499 3.1 10.2 0.5 500 -599 2.8 10.2 0.1 600 -699 1.2 4.6 0.0 700 -799 1.0 3.9 0.0 800 -899 1.5 5.6 0.0 900 -999 0.9 3.5 0.0 1,000 + 0.7 2.5 0.0
After excluding those hospitals and hospitals that were no longer running, there were a
total of 1,063 general hospitals and small hospitals as of September 3, 2008. This study
set these hospitals as the target population. There were 285 (26.8%) and 778 (73.2%)
general hospitals and small hospitals respectively (Table 2). Overall, there was an
average of 187.8 beds per hospital. The average number of beds in general hospitals was
411.7, with a maximum of 2,200 beds and a minimum of 100 beds. The average number
of beds in small hospitals was 105.8, with a maximum of 559 beds and minimum of 30
beds.
47
Table 3. Basic characteristics of hospitals in South Korea (foundation type) Hospital foundation type General hospitals Small hospitals All hospitals
N % N % N % Private-individual 66 23.2 582 74.7 648 61.0 Public 219 76.8 196 25.3 415 39.0 Total 285 100.0 778 100.0 1,063 100.0
There were 648 private-individual hospitals, 61.0% of the total hospital population; there
were 415 public hospitals, 39.0% of the population (Table 3).
Table 4. Basic characteristics of hospital in South Korea (geographic distribution) Location All hospitals General hospitals Small hospitals
N % N % N % N 1,063 100.0 285 100.0 778 100.0 State 01STATE (Seoul) 170 16.0 55 19.3 115 14.8 02STATE (Busan) 83 7.8 26 9.1 57 7.3 03STATE (Daegu) 80 7.5 9 3.2 71 9.1 04STATE (Daejeon) 22 2.1 7 2.5 15 1.9 05STATE (Ulsan) 30 2.8 3 1.1 27 3.5 06STATE (Gwangju) 43 4.0 18 6.3 25 3.2 07STATE (Incheon) 47 4.4 11 3.9 36 4.6 08STATE (Gyeonggi-do) 192 18.1 46 16.1 146 18.8 09STATE (Gangwon-do) 37 3.5 14 4.9 23 3.0 10STATE (Gyeongsangnam-do) 100 9.4 23 8.1 77 9.9 11STATE (Gyeongsangbuk-do) 56 5.3 16 5.6 40 5.1 12STATE (Chungcheongnam-do) 39 3.7 9 3.2 30 3.9 13STATE (Chungcheonbuk-do) 32 3.0 10 3.5 22 2.8 14STATE (Jeollanam-do) 68 6.4 18 6.3 50 6.4 15STATE (Jeollabuk-do) 55 5.2 14 4.9 41 5.3 16STATE (Jeju-do) 9 0.8 6 2.1 3 0.4 Urban or rural Rural 588 55.3 156 54.7 432 55.5 Urban 475 44.7 129 45.3 346 44.5
Many hospitals are located in or near Seoul, the capital of South Korea. There are 192
(18.1%) hospitals located in Gyeonggi-do, the state nearest to Seoul. There are 170
(16.0%) hospitals in Seoul. Nineteen percent of all general hospitals and 14.8% of all
small hospitals are located in Seoul. Approximately forty-five percent of hospitals are
located in urban areas and the rest, 55%, are located in rural areas (Table 4).
48
3.3. Definition of an EHR system
There is a clear difference between an EMR and an EHR. An EMR is an electronic
record of the patient’s clinical information electronically generated by encounters at one
particular healthcare provider, legally replacing a paper medical record. An EHR is an
electronic record of the patient’s healthcare information generated and aggregated by
one or more encounters of all of the care delivery organizations.104,105,106 The EHR is a
broader concept than the EMR. However, many scholars use the terms interchangeably
because the meanings are similar enough for the purposes of clinical research. For
clarity of expression, this study assumes that EHR includes EMR.
According to the academic field of computer science, a computer system is defined as a
system that consists of hardware and systems software working together to run various
application programs.107 Thus, this study descriptively defines an EHR system as an
electronic hardware and software system that creates, stores, and manages patient
clinical information by running various application programs in hospitals.
3.4. Data collection using survey instrument
This study developed a new survey instrument based on current relevant survey
instruments to address its hypotheses. The following sections describe the details of the
descriptive definitions and measures of each construct used in this study.
3.4.1. Measuring EHR adoption, use, and interoperability
EHR adoption: This study descriptively defines “adoption” of an EHR system as the
actual acquisition and implementation of an EHR system and operationally measures
49
EHR “adoption” as one of three levels: ‘fully implemented’, ‘partially implemented’, or
‘not implemented’. The study borrows questions on EHR adoption from the 2006
American Hospital Association’s IT survey43 and the 2005 IT survey in Korea.108 In
order to confirm their adoption status, this study follows Blumenthal’s (2006) idea of
asking hospitals when they installed their EHR system (i.e. introduction year) and how
they installed it (i.e. internally developed, outsourced, or purchased).109
For purposes of addressing the hypotheses, this study uses a binary variable: either
hospitals adopted an EHR system in any one of their clinical departments (e.g. hospital
ward or outpatient clinic) or they did not adopt and HER system in any department
(Table 5). The rationale is that hospitals with partial or full EHR adoption have a
common feature, which is the presence of an EHR system. This feature clearly
distinguishes hospitals that have not adopted an EHR system from those that have. The
objective of this study is to see how organizational and environmental factors affect
adoption. Therefore, partial and full adoption is grouped together as “adoption” because
they have similar characteristics which hospitals that have not adopted EHR do not have.
EHR use: This study descriptively defines “use” as the EHR functions that a hospital
has implemented, and this study operationally measured “use” as the number of reported
installed EHR functions. The study borrows questions on EHR use from the 2006
American Hospital Association’s IT survey, which included of 17 EHR functionalities.43
This study modifies and reduces the questions about EHR functionalities to 11 by
combining similar functions into single categories in order to simplify the questionnaire
and reduce redundancy (Table 5).
50
The functions this study uses are “access to current medical records (observations,
orders)”, “access to medical history”, “access to patient flow sheets”, “access to patient
demographics”, “order entry or results review (lab)”, “order entry or results review (any
radiology report)”, “order entry (real time drug interaction alerts)”, “other clinical alerts”,
“clinical guidelines and pathways”, “patient access to electronic records”, and “patient
support through home-monitoring (including self-testing, and interactive patient
education)”. This study asked the respondents whether these 11 functions were “fully
implemented,” “partially implemented” or “not implemented”. However, this study used
8 core functions in the final analysis excluding access function on medical records,
medical history, and patient demographics because they are general functions of EHR
systems.
These responses were transformed into a numeric scale for measuring hospitals’ EHR
use. The study gave a weighted value of 1.0 for full implementation, 0.5 for partial
implementation, and 0 for no implementation for each of the 8 functionalities. Then the
resulting values were summed to create a single numerical value for each hospital’s
EHR use. The numeric values ranged from zero to eight.
EHR Interoperability: This study descriptively defines “interoperability” as the
capacity of an EHR system to exchange healthcare information electronically with other
hospitals. Interoperability was measured based on the ability to electronically exchange
clinical information following HIMSS’s definition and NLM and HL7’s survey
instrument developed and used by Braithwaite’s, W.R., et. al. in 2005.52
51
Table 5. Measuring EHR adoption, use, and interoperability Construct Survey Question(s) Original Survey Measure Study
Scale Hypoth
esis*
EHR Adoption
Q5_1. Hospital wards EHR system 1. fully 2.partially implemented, 3. not implemented
Binary (anyone is
in 1,2, then
adoption=1, 0 in
otherwise)
HA 1-5
Q5_2. Outpatient EHR system 1. fully 2.partially implemented, 3. not implemented
EHR Use
Q6_1. Access to current medical records
0: no, 0.5: partial, 1: full use
Numeric (Sum of
all 11 items)
HU 1-5
Q6_2. Access to medical history Q6_3. Access to patient flow sheets Q6_4. Access to patient demographics Q6_5. Order entry or results review – Lab Q6_6. Order entry or results review – Rad. Q6_7. Order entry - Drug interaction alerts Q6_8. Other clinical alerts Q6_9. Clinical guidelines and pathways Q6_10. Patient access to electronic recordsQ6_11. Patient support (e.g., education)
EHR Interoperability
Q7_1. Patient history (family, patient)
1.fully, 2.partially exchanged, 3. not exchanged
Binary (anyone is
in 1,2, then
interoperability=1, 0
in otherwise)
HI 1-5
Q7_2. Patient summary (meds, etc.) Q7_3. Discharge summaries Q7_4. Laboratory information Q7_5. Imaging information Q7_6. Pharmacy information Q7_7. Others
* HA, HU, and HI mean all hypotheses related with EHR Adoption, Use, and Interoperability, respectively There were 22 capabilities in the original survey instrument, but this study modified and
reduced the number to 7 because the original instrument had too many irrelevant specific
survey items including public health and clinical trial areas. This study asked whether
the EHR system of respondents’ hospitals had the capability to transmit and receive 7
categories of clinical information electronically from other clinics or hospitals and
whether they were “fully exchanged”, “partially exchanged”, or “not exchanged”. The
seven categories this study used are patient history (family, patient), patient summary
(conditions, meds, and allergies), discharge summaries, laboratory information, imaging
information, pharmacy information, and other. These 7 categories are each categorized
52
as: fully exchanged, partially exchanged or not exchanged. However, the study finds that
many hospitals did not report any interoperability so the data distribution was heavily
centered on zero and one. Thus, this study transforms the numeric scale to a binary value
to address this issue; hospitals with no interoperability were given a score of zero, and
hospitals with any rating of either partial or full interoperability were given a score of
one (Table 5).
The rationale for this rescaling is that hospitals with partial or full EHR interoperability
have a common characteristic just like hospitals with full or partial adoption. Hospitals
should be grouped based on existence of interoperability in order to see how other
factors affect that feature. This categorization is also compatible with the definition of
EHR interoperability as the ability to perform healthcare information exchange with
other systems. Therefore, this study categorizes the respondents as either having
interoperability or not.
Finally, this study also considered information exchange methods. Among hospitals
reporting partially or fully exchanging patient’s information, this study only counts them
as having EHR interoperability if they use the following information exchange tools: e-
mail with attached document, internet access to patient records, a patient card system
such as IC card, health information technology system to system, and others. This is
because this study descriptively defines EHR interoperability as the capacity of an EHR
system to exchange healthcare information electronically with other hospitals like
mentioned above.
53
3.4.2. Measuring hospital internal features
The following section will describe the details of information on hospital internal
features, especially conceptual definition and its measurement in this survey instrument
starting with hospital basic characteristics.
Hospital basic characteristics: The variables considered as basic hospital
characteristics were foundation type, affiliation status such as multi-hospital system,
contracts for group purchasing and patient delivery, and urban or rural location. Of these
basic characteristics, this study only collects information on affiliation status and
contract status with other hospitals through this survey (Table 6). Affiliation status is yes
if the hospital is part of a multihospital system and no if the hospital is not part of a
system. The other information is collected through different information sources and
will be described in the next section.
Table 6. Hospital basic characteristics included in the survey Construct Question Measure Scale Hypothesis¹
Affiliation Q2. Multihospital system 1: Yes, 0: No Binary HA, HU, HI
Contract for group purchasing Q3. Multihospital system 1: Yes, 0: No Binary HA, HU, HI
¹ HA, HU, HI stands for hypotheses related with adoption, use, interoperability, respectively
Decentralization of decision-making: This study descriptively defines the de-
centralization of decision-making as the degree to which CIO or IT department
managers participate in important hospital decision-making. This is measured by the
reported degree of management participation of the CIO or IT managers in the hospital’s
decision-making regarding new employee hiring, current employee promotions, new
programs, and new policy, as suggested by Hage and Dewar (1973).63 This study uses a
five-level Likert scale to measure the degrees of management involvement. Each
54
gradation is given a numeric value (one to five) and the values for each question are
averaged to provide a final value to represent the hospital’s decentralization of decision-
making.
IT Staff specialization: This study descriptively defines IT staff specialization as a
hospital’s investment and support in human resources in the IT department and measures
the reported number of employees in charge of IT work in hospitals, which follows
Mintzberg’s perspective of viewing job specialization as “depth” to the control over an
organization’s work (p.69).110
This study initially asked the respondents to report the number of employees working in
the IT department with each of the following educational backgrounds: physician, nurse,
medical technician, management major, computer science major, etc. both before and
after EHR adoption. However, the survey results find that most employees are in the
category of computer science majors and there are no employees in most of the other
categories. To accommodate this finding this study combines all employees together
rather than separating them by specialty (Table 7). This study also finds that some
hospitals that did not have an IT department recorded their number of IT employees as
zero. In that case, this study allocates one IT employee to the hospital because all
hospitals have at least one person in charge of IT and EDI for billing and reimbursement
procedures in Korea. When the primary investigator contacted each hospital, this was
verified by the administrative managers in these hospitals.
To incorporate this measure into the hypothesis testing this study divides the number of
total IT employees by the number of beds and multiplied by 100. The rationale for
55
standardizing IT staff using bed size is because the number of IT employees increases as
the number of beds increases.
Table 7. Measures of factors affecting EHR adoption, use, and interoperability Construct Question Measure Scale Hypoth
esis³ Hospital structural factors
Decentralization of
Decision-making
Q9_a1. IT manager’s involvement: hiring 1. Never 2. Seldom 3. Sometimes 4. Often 5. Always
Numeric (average) H2, H4
Q9_a2. IT manager’s involvement: promotion Q9_a3. IT manager’s involvement: new rules Q9_a4. IT manager’s involvement: new policies
IT staff Specialization (Before EHR
adoption)
Q10_a1. No of IT staff majoring medicine Numeric Numeric (sum of all staff
from a1 to a6/bed *100)
H2, H4
Q10_a2. No of IT staff majoring nursing Numeric Q10_a3. No of IT staff majoring management Numeric Q10_a4. No of IT staff majoring medical technology Numeric Q10_a5. No of IT staff majoring computer science Numeric Q10_a6. No of IT staff with the other major Numeric
IT Infrastructure² (Before EHR
adoption)
Q4_a1. Existence of Outpatient CPOE¹ 0: no, 1: yes
Numeric (counting from a1 to
a8)
H2, H4
Q4_a2. Existence of Inpatient CPOE 0: no, 1: yes Q4_a3. Existence of Pharmacy dispensing system 0: no, 1: yes Q4_a4. Existence of Patient dis. processing system 0: no, 1: yes Q4_a5. Existence of Clinical laboratory system 0: no, 1: yes Q4_a6. Existence of Radiology department system 0: no, 1: yes Q4_a7. Existence of Intensive Care Unit system 0: no, 1: yes Q4_a8. Existence of Administrative procedures system 0: no, 1: yes
Organic structural form
Q12_1: Hierarchic structure of control 1-5 numeric
Numeric (average)
H2, H4
Q12_2: Emphasis on formal communication channels 1-5 numeric Q12_3: Strong insistence on a uniform managerial style 1-5 numeric Q12_4: Adopt true and tried management principles 1-5 numeric Q12_5: Strong emphasis on always getting personnel 1-5 numeric Q12_6: Tight formal control 1-5 numeric Q12_7: Emphasis on getting line and staff personnel 1-5 numeric
¹ Order Communication System (OCS) in Korea is almost the same term as Computerized Physician Order Entry (CPOE) System in the U.S. For the convenience of expression, this study interpreted OCS as CPOE. ² The system means whether each system is connected to hospital information system. ³ H2, H4 stands for hypotheses related with adoption (HA2, HA4), use (HU2, HU4), interoperability (HI2, HI4)
IT infrastructure: This study defines IT infrastructure as the physical and institutional
structures related to information technology needed for the operation of an organization.
IT infrastructure is measured by whether the following 8 areas are supported by the
Hospital Information Systems (HIS): outpatient computerized physician order entry
system (CPOE), inpatient CPOE, pharmacy drug management and dispensing system,
patient charging processing, clinical laboratory work, radiology work, intensive care unit,
56
and administrative procedures (Table 7). This study only counts the sub-systems which
existed before EHR adoption in order to see their impact on EHR adoption.
Organic hospital structure: This study descriptively defines organic hospital structure
as the degree to which a hospital emphasizes lateral communications, flexibility of rules
and regulation, and informal managerial control in its structure. On the contrary, this
study uses the term mechanical hospital structure to describe hospitals with fewer of
these features. Organic hospital structure is measured by whether the hospital had a
more mechanic structure or a more organic structure, following Burns & Stalker’s
(1961) construct. Questions on hospital structure are taken from Covin (1988)111 and
Kwon (2003)112 which were developed based on suggestions by Burns & Stalker (1961).
This study uses seven questions measuring the hospitals’ structure as either organic or
mechanic using a five-point Likert scale. Each gradation is assigned a numeric value
(one to five) and they are averaged across all items to provide a final value to represent
the hospital’s structure (Table 7).
3.4.3. Pilot survey
After developing the preliminary survey instrument, this study conducts a pilot survey of
five IT managers in Korea. There are several comments in the pilot survey. One
significant comment is that the term EMR was more widely used in Korea than EHR.
Other comments are that the survey needed to focus more on IT standards and to include
the respondent’s job title. Informants suggested several minor changes regarding the
translation of terms. Overall, respondents who participated in the pilot survey said that
57
there are few problems with the survey instrument. This final questionnaire used in the
study reflects all comments from respondents who participated in the pilot study.
3.4.4. Survey data collection
This study distributed the survey to IT departments in the Korean hospitals that made up
the target population. The survey respondents were chief information officers or IT
department managers. However, in hospitals with no IT department, this study allowed
the director of the administrative office to complete the survey because those directors
know when the hospital purchases new hardware and software programs. Moreover,
they are in charge of the electronic data interchange (EDI) process of requesting medical
claim reimbursement to the HIRA.
All survey data is collected through a website developed by the author. Before beginning
the development of the website and survey instrument, the author submitted this study’s
proposal to the Institutional Review Board (IRB) at the University of Minnesota (UMN)
and received an official letter from IRB on February 10, 2009. The official IRB response
letter reported that this study did not need approval because the subjects were hospitals
rather than human subjects.
After hearing back from the IRB, the author developed the website and survey
instruments for the study. The main homepage server was located at the UMN
(http://www.tc.umn.edu/~park0601/). However, since the UMN did not provide
application programs for database management such as SQL, this study used a local
server owned by a private company in Korea for all data storage. In order to prevent
58
unauthorized access to the survey instrument, this study created a password which
allowed only eligible persons to access the survey instrument.
On April 5, 2009, the author sent a one-time postal mail containing the access password
and a cover letter introducing the survey purpose and the website to the CIO or manager
of the IT department and the directors of administrative offices in each of the target
hospitals. This study assumes that the hospitals that participated in the survey agreed to
participate in the survey, and those who chose not to participate in the survey did not
agree to participate in the survey. In the letter, this study introduced the primary
investigator and explained the study purpose, data confidentiality, and participation
method. The hospitals’ mailing addresses were obtained from the KHA. During the
survey period, this study called or emailed the hospitals who participated in the survey to
inform them that their surveys were successfully completed on the website and to deliver
the author’s personal thanks for their participation.
To increase the response rate, this study directly contacted some hospitals’ IT
departments and administrative offices by calling the telephone numbers listed in the
booklet from the KHA. Since there were too many hospitals for the author to contact
individually, the author randomly selected two hospitals per local area, based on hospital
address in the alphabetical order and called one of them. If he was not able to get
through to the hospital staff, then the other hospital was contacted.
The Korea Information Technology of Hospital Association (KITHA) was contacted and
their support was obtained. Thus, this study was able to advertize and request
participation for this survey through their website (http://www.kitha.or.kr/index_1.asp).
The KITHA is a professional association composed of CIO and IT managers from 138
59
Korean hospitals. However, this study only contacted 106 hospitals because there were
many special hospitals such as veteran hospitals and long-term care hospitals that were
members of the KITHA. The KITHA also provided the author with the contact
information for the CIOs and IT managers of the member hospitals. After obtaining this
information, the author called each CIO and IT manager and sent e-mails requesting his
or her participation in the survey.
3.5. Additional data collection
The study needed to collect other information that was not part of the survey, such as
task complexity, market competition, number of beds, hospital types, and number of
hospitals in the local area. This study did not include these variables into the survey
because current government agencies and several professional associations already had
comprehensive datasets. Therefore, this information was gathered from other sources.
The following section describes the details of the descriptive definitions and measures of
constructs or confounding variables, and source of information used in this study starting
with hospital basic characteristics.
General characteristics: Foundation type was categorized as public or private hospital
(Table 8). The study used the term private hospital to describe individual or private
hospitals and the term public hospital to describe all other hospitals. This study
considered hospitals as urban if they were located in one of 6 metropolitan areas or
Seoul, and considered all other locations as rural. Information regarding the hospitals’
foundation types (private or public) and location was obtained from the KHA’s website
(http://www.kha.or.kr/).
60
Table 8. Hospital basic characteristics using other information sources Covariates Measure Scale Hypothesis¹ Source of Information
Foundation 1. Private, 2. Public Binary HA, HU, HI Korea Hospital Association
Location 1. Urban, 0. Rural Binary HA, HU, HI Korea Hospital Association ¹ HA, HU, HI stands for hypotheses related with adoption, use, interoperability, respectively
Task complexity: This study descriptively defined task complexity as the degree of
hospital task diversity and measured it by counting the number of specialty services the
hospital provides. Scott defined task complexity as the number of various inputs that
should be managed at the same time to produce outputs in organization.55 The average
number of specialties per hospital using the dataset from KHA was 13.1 with a
maximum of 27 and a minimum of 1 (Table 9). However, this study excluded five
medical specialties (tuberculosis medicine, nuclear medicine, emergency medicine,
preventive medicine, and industrial medicine) because small hospitals rarely, if ever,
offered these specialties.
Table 9. The number of specialties in study hospitals Number of specialties General hospitals Small hospitals ALL
N % N % N % Specialty distribution
1 - 3 0 0.0 27 18.2 27 7.6 4 - 6 0 0.0 54 36.5 54 15.2 7 - 9 10 4.8 52 35.1 62 17.4 10-12 30 14.4 13 8.8 43 12.1 13-15 39 18.8 2 1.4 41 11.5 16+ 129 62.0 0 0.0 129 36.2
N 208 148 356 Mean 18.0 6.1 13.1
Median 18.0 6.0 12.0 Max 27.0 14.0 27.0 Min 8.0 1.0 1.0 Std. 5.3 2.7 7.4
*: 27 medical specialties (internal medicine, pediatrics, neuroscience, neuropsychiatry, dermatology, surgery, thoracic surgery, orthopedics, neurosurgery, plastic surgery, obstetrics, ophthalmology, ENT (ear nose and throat), urology, tuberculosis, rehabilitation medicine, anesthesiology, diagnostic radiology, treatment radiology, clinical pathology, anatomical pathology, family medicine, nuclear medicine, Emergency medicine, occupational and environmental medicine, dentistry, and preventive medicine).
61
In addition in general hospitals and small hospitals, the minimum number of specialties
is determined by the number of beds. This is because Korea defines ‘general hospital’ as
medical facilities providing health care services with more than 100 beds and more than
7 specialties. Among those must be three of internal medicine, surgery, OBGN,
pediatrics, and the hospitals must also include imaging medicine, anesthesiology, and
laboratory medicine or pathology. In addition, the general hospitals must have at least 9
specialties if they have more than 300 beds. Korea defines ‘hospital’ (“small hospital” in
this paper) as medical facilities providing health care with more than 30 beds. Thus,
there is a high correlation among bed size, hospital type, and the number of specialties.
In order to reduce the multicollinearity introduced by these relationships, this study
excluded number of beds and hospital types (general hospitals or small hospital) from
the model. Information regarding the hospitals’ specialties was obtained from the
KHA’s website (http://www.kha.or.kr/).
Environmental complexity: This study defined environmental complexity as
complexity of environmental components influencing a focal hospital, such as diverse
healthcare providers, following Cannon and his colleagues’ perspective.113 It was
measured as: 1) the number of hospitals in a given geographic area, 2) true competitor
status based on a difference of the number of beds between any two hospitals. If the
difference between any two hospitals was less than 50 beds, this study counted both
hospitals as true competitors of each other in a given area, and 3) the Herfindahl-
Hirschman Index (HHI) (Table 10).114
Information on the number of beds in hospitals was obtained from the HIRA website
(http://www.hira.or.kr/cms/rb/rbb_english/index.html). This information was not used
62
directly in the study, but indirectly used for calculating HHI and identifying true
competitors and standardizing IT staff specialization. In case of environmental factors
such as number of hospitals within a local area, this study used local addresses to
determine locality.
Table 10. Measures of environmental complexity Construct Description Scale Hypothesis
Environmental Complexity 1 No. of hospital in a given area Numeric
HA3, HA5 HU3, HU5 HI3, HI5
Environmental Complexity 2 Existence of True competitor
Binary (1: true competitor, 0: no true competitor) (|d|=(bedi-bedj) bedi: bed size of hospital i if |d| <= 50 then hospitals i, j will be counted as 1; Else both i&j are counted as 0)
HA3, HA5 HU3, HU5 HI3, HI5
Environmental Complexity 3 Market competition
Numeric number
HHI¹=∑=
n
ii tbedbed
1
2)/(
n: the number of hospitals in a local area bedi: bed size of hospital i tbed: total bed in a local area
HA3, HA5 HU3, HU5 HI3, HI5
¹ The Hirschman- Herfindahl Index (HHI) measuring market competitions * All information came from other source (but not the survey)
3.6. Statistical data analysis
Study variables and their associations with EHR adoption, use, and interoperability are
first cross-tabulated and then tested one at a time using t-test of the mean difference or
non-parametric method, such as Wilcoxon’s rank sum test. Overall, this study uses
Generalized Estimating Equations (GEE), an extension of the Generalized Linear Model
(GLM),115 and General Linear Mixed Model (GLMM) to examine the proposed
hypotheses.
Two things critically affect why this study chooses GEE and GLMM. First, this study
has a hierarchical (multi-level) or cluster correlated data. Each hospital is nested within
63
an environment (or a local area). For example, the environmental variables of the
hospitals within a local area have the same values, such as number of hospitals
(environmental complexity measure). Since they can equally affect other hospital
constructs within the same local area, this kind of dataset inevitably causes a correlation
issue within the same cluster and, thus, needs different statistical approaches. Second,
this study has three outcome variables with two different scales of outcome measures:
one (the use of EHR) is numeric values and the two others (adoption and
interoperability) are a binary value. GEE and GLMM can resolve issues caused by both
the correlation concern and the two scales of outcome variables.
The method of GEE is indeed frequently used to analyze correlated datasets having
hierarchical, nested, or longitudinal data structure, especially if response variables are on
a binary scale.116 GLMM is a general statistical approach dealing with the correlation
issue arising in a hierarchical data structure, especially if the scale of the outcome
variables is numeric.117 The next section discusses specific statistical considerations for
the data analysis of two different outcome variables.
Finally, there are several reasons that this study uses GEE and GLMM, but not Survival
Analysis or Path Analysis, as the final analytical methods. This study uses retrospective
cross sectional survey data and, thus, it does not fit well in Survival Analysis which
requires a longitudinal data structure. Some variables of this study have longitudinal data
structure, but the majority of the variables are measured in a cross-sectional fashion.
Path Analysis can be another alternative for the final data analysis because it is a kind of
extension of regression analysis.118 However, this study adopts a regression model
because this study uses several binary variables in a hospital’s general characteristics
64
and hierarchical data structure, which is described in the following paragraph. Path
analysis does not resolve these issues yet.
3.6.1. Binary outcome variables: EHR adoption and interoperability
There are also two different analytical approaches when we have a correlated dataset
such as cluster, hierarchical, or longitudinal structure and its outcome variable is binary:
marginal-effect models and random-effect models.119, 120 The former method uses GEE.
It is sometimes called population-averaged models.121 This can be achieved using the
GENMOD procedure with a logistic link function and the repeated-statement in SAS.
The latter one, random-effects models, is called generalized linear mixed models or
multilevel models119 and can be analyzed using the NLMIXED procedure with random
statement in SAS. Both are appropriate analytical methods for this study because they
deal with correlation issues arising from environmental factors in this study and binary
outcome variables.
The following describes how marginal-effect models and random-effect models differ
and why this study chooses marginal-effect models using GEE. Suppose Yij has only
two possibilities (Yij = 1: EHR adoption, Yij = 0: no EHR adoption) and Xij is a design
matrix of hospital covariates with 1xK vector and β0, β1 are intercept and Kx1 vectors of
estimates of covariates where the j-th hospital is located at the i-th environment (i=1 to n,
j=1 to m). Thus, the expected EHR adoption rate equals pij=P(Yij=1). A marginal logistic
regression model is formulated120 by:
Logit(pij) = β0 + β1Xij, Var(Yij) =pij(1- pij), where Corr(Yij, Xik) = α
α: a constant or a working correlation matrix
65
In this model we assume that any correlation between two hospitals from the same
environment is identical and any correlation between two hospitals located in different
environments is considered independent because of clustered correlated
characteristics.120 Thus, we need to adjust for the correlation between hospitals from the
same environments. By specifying a correlation structure, the model reflects the non-
independence of observations.115 This means that a group effect is not explicitly
expressed in the model, but the correlation of the measurements is explained through a
covariance matrix.121 The interpretation of the regression coefficients is affected by the
actual averages of the observation effects of the variables across the environments. 120
In contrast, a random effect model assumes that heterogeneity exists across the cluster
which in this case is environment. A random effect model is expressed120 by:
Logit(P(Yij = 1) | ui) =β0 + β1 Xij + ui, with ui ~ N(0, σ²).
This model explicitly models heterogeneity with the subject specific environmental
probability.121 By introducing a random effect term, ui where the variance σ² measures
the degree of heterogeneity in the probability of EHR adoption, and the model can get
the regression coefficient adjusting for the correlation from environments.120 Thus,
marginal model differs from random-effect model, but they reflect the correlation issues
into the model in a different way.
This study chooses marginal-effect models using GEE for several reasons. First, the
focus of this study is to predict and use the average odds ratios of the two hospital
groups, adopting and not adopting an EHR system and having and not having EHR
interoperability, but not the individual hospital’s odds ratios. Second, explicitly
66
accounting heterogeneity from environment subject to subject is not the chief interest of
this study. This study only needs to adjust for the correlation between hospitals within
the same local area to investigate the relationship between outcome variable and
proposed constructs. In both cases, the marginal-effect model using GEE is
recommended over the random-effect model in literature,121,115 and, thus, this study
selects the marginal-effect model. The interpretation of the regression coefficient of a
marginal effect model is analogous to the standard logistic regression.
3.6.2. Numeric outcome variable: EHR use
With respect to the use of EHR outcome variables, this study used a GLMM and the
specific method of analysis was as follows.
Model explanation: this study has a nested data structure. Hospitals are nested within
each environment. Hospitals within the same environment may be correlated to each
other. General Linear Mixed Model (GLMM), with fixed effects and random effects in
the same model, takes into account the variance-covariance structure to produce more
efficient coefficient estimates. A simple linear regression model is:
Yij = α0+α1*Xij+εij ………. (E1)
Yij represents the outcome variable for the j-th hospital within the i-th environment, α0 is
the intercept, α1 is a vector of fixed effects with design matrix Xij, and εij is an unknown
random error. The GLMM extends the simple model to a model with a random effect
67
parameter specific to environment and an unknown random error. The mathematical
notation of GLMM is:
Yij = α0+α1*Xij+βi+εij ……(E2)
The model shows the fixed effect component α1Xij (contribution of independent
variables to the variation of dependent variables) and the random effects component βi (a
correlation among outcomes within environment). According to GLMM, βi ~ N (0,
σenvir.²) and εij ~ N (0, σe²), Var (Yik) = σenvir.²+σe² and Cov (Yij, Yik) = σenvir². Thus, Corr
(Yij, Yik) = σenvir.²/ (σenvir.² +σe²), which is constant. This means that the variance of any
hospital within an environment (local area) is equal to σenvir.² +σe², and the covariance of
any two hospitals within an environment (or local area) is equal to σenvir.². This
corresponds to a model with compound symmetry (or exchangeable) variance-
covariance structure in a general linear model (GLM).
Main Effect of task, structure, and environment: the main effects of task, structure,
and environment on the use of EHR were examined without any interaction terms. To
determine whether the environment affects the use of EHR, this study tested whether
there is a random intercept effect in the model.
Full model: Yij = α0+α1*X1+α2*X2+α3*X3+α4*X4+βi+εij ……....(E3)
Reduced model: Yij = α0+α1*X1+α2*X2+α3*X3+α4*X4+εij …….(E4)
Where i=environment (local area) and j = hospital within each local area.
ijY =the use of EHR (or interoperability) of hospital j at local area i,
X1= the vector of task variables
X2= the vector of structural variables
X3= the vector of environment characteristics
68
X4= the other hospital covariates (e.g., bed)
βi = environment-specific random intercept for environment i
εij = random error term for hospital j in environment i
According to statistical theroy, βi ~ N (0, σenvir.²), εij ~ N (0, σe²), and if what we are
testing is a hypothesis: H0: σenvir.² = 0 versus Ha: σenvir.² > 0. If σenvir.² = 0 after testing
likelihood ratio test, then βi is constant. This means that we need a reduced model (E4).
If not, we need a random effect model (E3).
Analysis of any interaction effects: to see the interaction effects between each
construct and the use of EHR, this study included an interaction term into the model
with the main variable. For example, with respect to HU4, the interaction terms between
task and four structural variables were added to the model:
(1) If there is no environmental random effect, then the equation would be:
Yij =α0+α1*X1+α2*X2+α3*X3+α4*X4+α5*X1*X2+εij ………….(E5)
(2) If there is significant environmental random effect, then the equation would be:
Yij = α0+ α1*X1+α2*X2+α3*X3+ α4*X4+ α5*X1*X2+βi+εij …...(E6)
In the model, α5*X1*X2 represents the interaction terms between task and four structural
variables. Although there are significant main effects for task and any of the structural
variables, the interaction between task and each of the structural variables was tested in
the model.
69
3.6.3. Handling multicollinearity issue
A multicollinearity issue occurs when the model incorporates highly correlated
independent variables. This issue leads to an increased variance of parameter estimates,
which reduces the significance of individual independent variables. This occurs even
though the overall model’s p-value is very low. Thus, it would result in a biased estimate
of the regression coefficient without resolving multicollinearity issues. For those reasons,
this study examined the correlation between independent variables before fitting them
into the model. If some variables are highly correlated to each other (r>0.6), then this
study eliminated one of those variables.
3.6.4. Procedures for any interaction terms of regression model
Several scholars on statistics and related studies have recommended mean centering
when researchers have any interaction terms of numeric variables in the regression
models.122,123 Mean centering means subtracting the mean of a variable of a study
subject from all observations of the variable and using the subtracted value in the
regression model. There are two main reasons for recommending this procedure.
First, including two variables with multiplication into the model may cause multi-
collinearly issues. Researchers may avoid this issue by using mean centering in the
model. Second, logical and easy interpretation of the regression coefficient of the
interaction term is available because researchers can assume appropriate levels of
boundary of the mean centered variable. Thus, this study will use mean centered values
only if the proposed model has any interaction terms.
70
For the visual representation of any interaction terms of the proposed models using
graphs, this study used SAS (version 9.2) and the website of Academic Technology
Services at the University of California in Los Angeles (UCLA).124 This study
downloaded their initial SAS codes on visualizing interactions and modified those codes
for this study’s purpose.
3.7. Reliability test on survey instruments
This study used two important constructs that were subjectively measured:
decentralization of decision-making and organization structure. The former was
measured using four questions which were asked both before and after EHR adoption.
The latter was measured by seven questions asked only after EHR adoption. This study
looked at the reliability of those variables using the Cronbach’s alpha. The Cronbach's
alpha measures the internal consistency of reliability, which is how well a group of
independent variables measures a construct.125 The Cronbach’s alpha of the constructs
measuring decentralization of decision-making in the past and present were 0.870 and
0.861 respectively. The Cronbach’s alpha for organizational structure was 0.716 (Table
11).
Table 11. Reliability of subjective survey questions Constructs Initial no. of
items Final no. of
items Cronbach’s
alpha Decentralization of decision-making (past)¹ 4 4 0.873 Decentralization of decision-making (current) 4 4 0.847 Organizational structure 7 7 0.718
¹ Only asked to the hospitals adopting EHR systems
The recommended level of Cronbach’s alpha differs between scholars, but Nunnally sets
forth the recommended level as 0.7 in the case of preliminary research.126 Kwon, citing
71
Van de Ven and Ferry (1980), noted that having more than 0.6 of Cronbach’s alpha
would be enough in a case of exploratory research or if the unit of analysis is not the
individual, but the organization (Kwon, 2003). This study is exploratory study and the
unit of analysis is organization. Thus, this study argues that Cronbach’s alpha is high
enough for this study, and included all questions in the analysis.
3.8. Correlation analysis among the variables
The purpose of this study is to investigate the impact of hospitals’ internal features, such
as task, structure, and environmental factors on EHR adoption, use, and interoperability.
To analyze the relationship using multivariate regression, the relationship among
predictor variables should be independent because some dependent relationships result
in multicollinearity, which affects significance and interpretation of the regression
coefficients. In order to prevent this problem, this study investigated the relationship
among independent variables and excluded the variables that had high correlation and
were less important from the model.
Table 12 shows the Spearman’s correlation among dependent and independent variables.
Compared to the Pearson’s correlation, measuring the linear relationship between two
variables when the variables have an equal-appearing intervals scale, Spearman’s
correlation is calculated based on using non-parametric methods of association.127 This
study used Spearman’s correlation to check multicollinearity issues among the variables
because some of the variables in general characteristics were measured with a binary
scale and the distribution of some variables could not be assumed as normality. Among
72
the variable, HHI was highly negatively correlated with the number of hospitals within a
local area (r=-88). Bed size was also highly correlated with the number of medical
specialties. Thus, this study excluded these variables from the final regression model.
Table 12. Correlations between variables of all study hospitals Spearman Correlation Coefficients, N = 352
Prob > |r| under H0: Rho=0 Adoptio
n Bed Foundati
on Multihos
pital Contract Urban Medical
SpecialtyDecentralization
Specialization
IT infrastru
cture
Organic structure
True competit
or
Hospital withinarea
HHI
Adoption 1.00
Bed 0.11 1.00 0.05
Foundation 0.06 -0.51 1.00 0.30 <.0001
Multihospital System
0.04 0.28 -0.48 1.00 0.47 <.0001 <.0001
Contract for Group purchasing
0.15 0.10 -0.06 0.19 1.00 0.01 0.06 0.24 0.00
Urban (vs. Rural)
0.08 0.05 0.07 -0.04 -0.01 1.00 0.12 0.39 0.18 0.46 0.81
Medical Specialty 0.12 0.85 -0.53 0.27 0.13 0.07 1.00 0.03 <.0001 <.0001 <.0001 0.01 0.20
Decentralization 0.13 0.15 -0.04 0.05 0.03 0.07 0.19 1.00 0.01 0.00 0.45 0.33 0.60 0.21 0.00
Specialization 0.01 -0.13 0.01 0.00 0.05 0.15 0.05 0.25 1.00 0.92 0.02 0.86 1.00 0.35 0.00 0.35 <.0001
IT infrastructure 0.45 0.38 -0.27 0.14 0.22 0.01 0.42 0.03 -0.05 1.00 <.0001 <.0001 <.0001 0.01 <.0001 0.85 <.0001 0.63 0.35
Organic structure 0.18 -0.07 0.12 0.04 0.04 -0.04 -0.10 0.01 0.00 0.01 1.00 0.00 0.18 0.03 0.49 0.45 0.43 0.05 0.88 0.99 0.83
True competitor -0.04 -0.55 0.38 -0.20 -0.12 0.04 -0.45 -0.07 0.06 -0.16 0.08 1.00 0.47 <.0001 <.0001 0.00 0.03 0.46 <.0001 0.19 0.29 0.00 0.16
Hospitals within Local area
0.10 0.03 0.07 -0.07 0.07 0.15 0.02 0.05 0.19 0.06 -0.01 0.30 1.000.07 0.62 0.17 0.20 0.18 0.01 0.67 0.37 0.00 0.24 0.82 <.0001
HHI -0.04 0.04 -0.11 0.12 -0.05 -0.06 0.06 0.02 -0.06 -0.06 -0.03 -0.40 -0.88 1.000.41 0.41 0.04 0.03 0.38 0.27 0.24 0.67 0.25 0.28 0.57 <.0001 <.0001
Table 13 shows the relationship among EHR use, EHR interoperability, and other
variables. With respect to the analysis of EHR use and interoperability, this study only
selected hospitals that adopted an EHR system and checked correlation among the
variables. There were 158 hospitals with an EHR system. The HHI still had high
correlation with the number of hospitals within a local area (r=-0.85) and, thus, this
study excluded this variable from the regression analysis, but included the number of
hospitals within a local area.
73
Table 13. Correlations between variables of hospitals only having EHR adoption Spearman Correlation Coefficients, N = 158
Prob > |r| under H0: Rho=0 USE Interope
rability Foundati
on Multihospital sys.
Contract Urban Specialty Decentralization
Specialization
IT Infrastru
cture
Organic Structur
e
True Competit
or
HospitalsWithin
area
HHI
EHR USE 1.00
Interoperability 0.18 1.00 0.02
Foundation -0.02 -0.12 1.00 0.78 0.14
Multihospital System
0.22 0.09 -0.55 1.00 0.00 0.29 <.0001
Contracts 0.19 0.20 0.02 0.18 1.00 0.02 0.01 0.83 0.03
Urban (vs Rural)
-0.02 0.03 0.03 -0.03 -0.10 1.00 0.83 0.69 0.74 0.76 0.22
Specialty (task Complexity)
0.13 0.27 -0.57 0.41 0.20 0.04 1.00 0.10 0.00 <.0001 <.0001 0.01 0.62
Decentralization
0.00 0.06 -0.14 0.09 0.04 0.03 0.32 1.00 0.98 0.43 0.08 0.28 0.66 0.68 <.0001
Specialization 0.03 0.13 -0.11 0.15 0.05 0.08 0.15 0.28 1.00 0.72 0.10 0.15 0.06 0.57 0.32 0.06 0.00
Infrastructure 0.09 0.01 -0.14 0.20 0.07 -0.10 0.11 -0.08 -0.05 1.00 0.24 0.88 0.08 0.01 0.35 0.20 0.18 0.35 0.55
Organic structure
0.00 0.04 0.15 -0.02 -0.04 -0.07 -0.10 -0.13 -0.06 -0.01 1.00 0.95 0.63 0.06 0.76 0.61 0.41 0.20 0.12 0.43 0.91
True competitor
-0.07 -0.16 0.42 -0.23 -0.07 0.01 -0.56 -0.10 0.02 -0.08 0.07 1.00 0.39 0.05 <.0001 0.00 0.37 0.89 <.0001 0.22 0.77 0.31 0.35
Hospitals within Local area
0.02 -0.02 0.17 -0.07 0.04 0.09 -0.10 0.11 0.19 -0.08 -0.07 0.28 1.000.84 0.77 0.03 0.40 0.61 0.28 0.20 0.16 0.02 0.29 0.41 0.00
HHI 0.02 0.06 -0.24 0.14 -0.02 0.01 0.21 -0.02 -0.01 0.06 -0.04 -0.42 -0.85 1.000.82 0.46 0.00 0.08 0.78 0.87 0.01 0.78 0.88 0.45 0.59 <.0001 <.0001
74
CHAPTER IV RESULTS
4.1. General characteristics of study hospitals
Table 14 shows the general characteristics of the study hospitals. A total of 356 hospitals
responded to the survey, which was a 33.5 percent response rate from the 1,063 total
hospitals. Four hospitals were excluded from the analysis because they adopted an EHR
system at the time of their establishment and this study needed information on hospital
status before adopting an EHR system. Thus, a total of 352 hospitals were analyzed for
this study.
Table 14. Basic characteristics of study hospitals Basic characteristics All hospitals
Study hospitals analyzed 352.0 Response rate (%) ¹ 33.5 % of private hospitals 40.6 % of multihospital systems 46.6 % of contract for patient delivery or group purchasing 83.2 % of urban location 88.4 State location
01STATE (Seoul) 19.0 02STATE (Busan) 8.5 03STATE (Daegu) 5.4 04STATE (Daejeon) 1.7 05STATE (Ulsan) 3.1 06STATE (Gwangju) 5.1 07STATE (Incheon) 3.1 08STATE (Gyeonggi-do) 11.7 09STATE (Gangwon-do) 4.8 10STATE (Gyeongsangnam-do) 9.7 11STATE (Gyeongsangbuk-do) 5.4 12STATE (Chungcheongnam-do) 4.6 13STATE (Chungcheonbuk-do) 3.1 14STATE (Jeollanam-do) 8.2 15STATE (Jeollabuk-do) 4.8 16STATE (Jeju-do) 1.7
¹Total hospital population=1,063, responding hospitals=356. This study excluded 4 cases having an EHR system with their establishment.
75
There were 143 private-individual hospitals (40.6%) and 164 hospitals (46.6%)
maintaining a multi-hospital system. Eighty-three percent of the study subjects had
contracts with other hospitals for group purchasing and patient delivery. Nineteen
percent (67) and 11.7 percent (41) of study hospitals were located in Seoul and
Gyeonggi-do province, respectively.
4.2. EHR adoption
4.2.1. Hospital characteristics by EHR adoption
This section describes the general characteristics and structural features by an EHR
adoption status (Table 15). Forty-five percent of the study hospitals (158) reported
adopting an EHR system. Hospitals reporting the contract with other hospitals for patient
delivery and group purchasing had higher EHR adoption rates than hospitals without any
contracts (48.1% vs. 28.8%, p=0.0065). When defining urban and rural area based on
metropolitan cities, EHR adoption rate was higher in hospitals located in urban areas
than hospitals in rural areas (47.8% vs. 26.8%, p=0.0134).
Table 15. Basic characteristics of study hospitals by EHR adoption status Basic characteristics Adopting EHR
system Not adopting
EHR N p-value ¹
% of an EHR system adoption status 44.9 55.1 352(100.0) -
Foundation Private 48.3 51.7 143 (100.0)
0.2937 Public 42.6 57.4 209 (100.0)
Multihospital systems Yes 47.0 53.1 164 (100.0)
0.4669 No 43.1 56.9 188 (100.0) Contract for group purchasing
Yes 48.1 51.9 293 (100.0) 0.0065 No 28.8 71.2 59 (100.0)
Location Urban 47.3 52.7 311 (100.0) 0.0134 Rural 26.8 73.2 41 (100.0)
¹ Chi-square test
76
Table 16 presents hospitals’ internal features by EHR adoption status. Task complexity
of hospitals was significantly related to EHR system adoption (p=0.0293). The table also
shows the CIO or IT manager’s participation in decision-making by EHR adoption
status. This CIO or IT manager’s participation in decision-making was measured with a
five-point Likert scale; a higher number means that IT managers participate more
actively in the hospital’s overall decision-making processes. Hereafter it will be labeled
a de-centralized decision-making system when IT managers do actively participate and a
centralized decision-making system when they do not. The average score of four
measures of decentralization of decision-making in hospitals with an EHR system before
they adopted the system was significantly higher than those of hospitals who have not
adopted an EHR system, 3.20 and 2.91, respectively (p=0.0137). The score was
significantly higher for hiring (p=0.0330) and promotion (p=0.0030) for hospitals that
have adopted an EHR system.
Table 16. Hospital’s internal features by EHR adoption status Basic characteristics Adopting
EHR systemNot adopting
EHR N (%) p-value†
Total 158 194 352 - Task # of medical specialties¹ 13.12 11.62 352 0.0293
Structure
Decision making (average) 3.20 2.91 352 0.0137 Hiring 2.99 2.63 352 0.0330 Promotion 2.72 2.22 352 0.0034 Program 3.93 3.87 352 0.2456 Policy 3.15 2.93 352 0.1611
IT staff specialization³ 1.69 1.56 352 0.9157 IT infra-structure 7.87 6.45 352 <0.0001 Organic structure 2.74 2.48 352 0.0009
Environment
Having true competitors Yes 43.22 56.78 199 (100.0) 0.4724 No 47.06 52.94 153 (100.0)
HHI score 0.30 0.32 352 0.4128 # of hospitals within the area 8.83 7.64 352 0.0712
¹ # of the medical specialties adjusting for the type of hospitals and bed size (see, research method section) ² A 10-standards item used (yes: if there is at least one standardized IT item, no: otherwise) ³ Standardized using bed size: the number of IT staff/bed*100 † p-value of Wilcoxon’s Rank Sum test or chi-square test
77
The number of IT infrastructure items was significantly higher in hospitals with an EHR
system (7.87) than in hospitals without an EHR system (6.45) (p<0.0001). This study
also examined whether hospitals with an organic management structure were more likely
to adopt EHR systems following Burns and Stalker’s definition (1961). Their arguments
were strongly supported in this case. Organic structure was measured on a Likert scale
with most mechanical (least organic) structure having a value of 1 and the most organic
structure having a value of 5. Thus, a high number means that the hospital had a more
organic structure. The average scores of hospitals with an EHR system were
significantly more organic (2.74) than those without an EHR system (2.48) (p=0.0009).
The average number of hospitals within the local area of hospitals with an EHR system
was marginally higher than those of hospitals without an EHR system, which were 8.83
and 7.64 respectively (p=0.0712).
4.2.2. EHR adoption and task complexity
This section describes the results of the logistic regression analyzing the relationship
between task complexity and EHR system adoption before (Model 1) and after (Model
2) controlling for hospital covariates.
Hypothesis Addressed:
HA1: Hospitals with high task complexity were more likely to adopt an EHR system.
For a 1-unit increase in task complexity measured by the number of medical specialties,
the odds of EHR adoption were estimated to increase by a multiplicative factor of 1.062
78
after controlling for hospital covariates. Among hospital covariates, hospitals with
private foundation (versus public) and with contracts for patient delivery, group
purchasing (versus not having any contract) were significantly more likely to have
higher odds of adopting an EHR system (Table 17).
Table 17. Impact of task complexity on EHR adoption
Independent variables Model 1 (Odds ratio)
Model 2 (Odds ratio)
Foundation type (private) (ref=public) 2.261* Multi-hospital system (yes) (ref=no) 1.320 Contracts with other hospitals (group purchasing, etc.)(ref=no) 2.181* Location (urban) (ref=rural) 1.284 Task Complexity (# of specialties) 1.036* 1.062*
* < 0.05, ** < 0.005
4.2.3. EHR adoption and hospital structure
This section describes the results of the relationship between EHR adoption and hospital
structure before (Model 1) and after (Model 2) controlling for hospital covariates.
Hypothesis Addressed:
HA2: Hospitals with decentralized decision making, high IT staff specialization, high
IT infrastructure, and an organic managerial structure were more likely to adopt an
EHR system.
A one-unit increase in IT infrastructural items leads to an increase by a factor of 3.123 in
the odds of adopting an EHR system after controlling hospital covariates. The predicted
odds of adopting an EHR system increase to 1.717 times multiplicatively for a 1-unit
increase in the score of hospital organic form (Table 18).
79
Table 18. Impact of hospital structure on EHR adoption Independent variables Model 1
(Odds ratio) Model 2
(Odds ratio) Foundation type (private) (ref=public) 2.258** Multi-hospital system (yes) (ref=no) 1.168 Contracts with other hospitals (ref=no) 1.494 Location (urban) (ref=rural) 1.347
Structure before EHR adoption
Decentralization of decision-making system 1.259* 1.242 IT staff specialization 1.078 1.039 IT infra-structure 2.901* 3.123* Organic structure 1.871** 1.717**
* < 0.05, ** < 0.005
4.2.4. EHR adoption and environment
This section investigates the relationship between EHR adoption and environmental
complexity before (Model 1) and after (Model 2) controlling for hospital covariates
(Table 19). This study selected three variables representing environmental complexities:
HHI, true competitor, and the number of hospitals within the local area. However, as
mentioned in the research method section, there was a high correlation between HHI and
the number of hospitals within a local area so the model only includes the number of
hospitals within the local area. In order to define the local area, this study used
administrative district which is equivalent to counties in the United States.
Hypothesis Addressed:
HA3: Hospitals with high environmental complexity were more likely to adopt an EHR
system.
Environmental complexity significantly impacted EHR adoption before and after
controlling for hospital covariates. After controlling for hospital covariates, the odds of
80
EHR adoption were estimated to be increased by 4.0% for a unit increase in the number
of hospitals within a local area.
Table 19. Impact of environmental complexity on EHR adoption
Independent variables Model 1 (Odds Ratio)
Model 2 (Odds Ratio)
Foundation type (private) (ref=public) 1.419 Multi-hospital system (yes) (ref=no) 1.348 Contracts with other hospitals (group purchasing, etc.)(ref=no) 2.360* Location (urban) (ref=rural) 1.376* Environmental complexity (All hosp. within the local area) 1.040* 1.040**
* < 0.05, ** < 0.005 4.2.5. EHR adoption, task, structure, and environment
Traditional contingency theory argues that organizations that have more complex tasks
are more likely to decentralize their decision-making systems. If this argument is true,
then there should be a relationship between EHR adoption and the four constructs of the
hospital structure measure. This study examined if these kinds of arguments are
supported in hospitals adopting EHRs by testing the following two hypotheses.
Hypothesis Addressed:
HA4: The relationship between task complexity and an EHR system adoption were
dependent on hospital structure.
HA5: The relationship between hospital structure and an EHR system adoption were
dependent on environmental complexity.
Table 20 presents the results of analyses investigating whether hospital structural factors
work as moderators between task complexity and EHR adoption and between
environmental complexity and EHR adoption. The results show that hospital structural
81
factors had no significant moderating effect between environmental complexity and
EHR adoption, but did have some significant effects between task complexity and an
EHR system adoption.
Table 20. Impact of a hospital’s internal features on EHR adoption Independent variables Odds ratio
Intercept 0.230** Foundation type (private) (ref=public) 2.356** Multi-hospital system (yes) (ref=no) 1.117 Contracts with other hospitals (group purchasing, etc.)(ref=no) 1.443 Location (urban) (ref=rural) 1.232 Task complexity # of specialties 1.026
Structure before EHR adoption
Decentralization 1.199 IT staff Specialization 1.132 IT infra-structure 2.820* Organic structure 1.656**
Envir. Complexity All hosp. within area 1.059 Task complexity*Structure
Task*Decentralization 1.059** Task*IT staff Specialization 1.010 Task*IT infrastructure 0.952* Task*Organic structure 1.010
Environmental complexity*Structure All hosp. within area*Decentralization 1.014 All hosp. within area*Specialization 0.993 All hosp. within area*IT infrastructure 0.960 All hosp. within area*Organic structure 0.984
* < 0.05, ** < 0.005, Note: all numeric values were analyzed using mean centered predictors For the interaction effect between task complexity measured by the number of medical
specialties in a hospital and decentralization of decision-making score of IT managers or
CIOs, the impact of task complexity on the odds of adopting an EHR system critically
relies on decentralization of decision-making system. The logistic regression models the
log odds of a response (a probability of an event versus non-event) as a linear
combination of the predictors.128
Using this linear relationship, Figures 7 and 8 illustrate the association among task
complexity, decentralization of decision-making score, and the log odds of adopting an
EHR system. Under a highly decentralized decision-making system, which is the right
82
side of its mean centering points (X2=0), the log odds of adopting an EHR system
increase as task complexity increases (Figure 7).
Figure 7. Task complexity, decision-making structure, EHR adoptions 1
Figure 8. Task complexity, decision-making structure, EHR adoptions 2
83
However, under a highly centralized decision-making system which is moving to the
right side from its mean centering points (X2=0), the log odds of adopting an EHR
system decrease when task complexity increases (Figure 8).
With respect to the interaction effect between task complexity and IT infrastructure, the
impact of task complexity on the odds of adopting an EHR system is dependent on the
level of IT infrastructure. When we observe a moving direction from its mean centering
points (X2=0), the log odds of adopting an EHR system in hospitals decreases as task
complexity increases under a high level of IT infrastructure which is above the average
of the group (Figure 9).
Figure 9. Task complexity, IT infrastructure, EHR adoptions 1
However, compared to this direction, the log odds of the adoption of an EHR system
slightly increases as task complexity increases under a lower level of IT infrastructure
which is below the average of the group (Figure 10).
84
Figure 10. Task complexity, IT infrastructure, EHR adoptions 2
4.3. EHR use
This section investigates the relationships among a hospital’s internal characteristics,
such as task, structure, and environment, and EHR use using relevant statistical methods.
4.3.1. Hospital characteristics by EHR use
There were 158 hospitals that adopted an EHR system among the 352 hospitals that
responded to the survey. Hospitals maintaining multihospital systems and that have a
contract with other hospitals were using more EHR functions than single hospitals
without any contract (5.69 versus 4.95, p=0.0052; 5.43 versus 4.29, p=0.0162) (Table
21).
85
Table 21. Characteristics of hospitals adopting an EHR system by EHR use Basic characteristics EHR use p-value¹
N 158
Foundation Private 5.25 0.7754 Public 5.36
Multihospital systems Yes 5.69 0.0052 No 4.95
Contract for patient delivery Yes 5.43
0.0162 No 4.29
Location Urban 5.29 0.7032 Rural 5.59 ¹ p-value of Wilcoxon’s Rank Sum test, but Kruskal-Wallis Test used for bed size
4.3.2. EHR use and task complexity
This study investigated the impacts of task complexity on EHR use before (Model 1)
and after (Model 2) controlling for a hospital’s general characteristics.
Hypothesis Addressed:
HU1: Hospitals with high task complexity were more likely to use an EHR system.
The result showed that task complexity did not have any significant effect on EHR use
(Table 22). Hospitals maintaining a multihospital system had one more EHR function
than the single hospital groups.
Table 22. Impact of task complexity on EHR use
Independent variables Model 1 (Beta¹)
Model 2 (Beta)
Foundation type (private) (ref=public) 0.347 Multi-hospital system (yes) (ref=no) 0.795* Contracts with other hospitals (group purchasing, etc.)(ref=no) 0.889 Location (urban) (ref=rural) 0.217 Task Complexity (# of specialties) 0.021 0.006
* < 0.05, ** < 0.005, ¹ Beta stands for regression coefficient
86
4.3.3. EHR use and structure
This section investigates the relationship between EHR use and the four factors for
hospital structure: (1) de-centralization of decision-making of the IT department on
hospital operations such as hiring, promotion, changing programs, and policies, (2) IT
staff specialization, (3) IT infrastructure, and (4) organic structural form.
Hypothesis Addressed:
HU2: Hospitals with decentralized decision making, high IT staff specialization, high
IT infrastructure, and an organic managerial structure were more likely to use an EHR
system.
Table 23 shows the results from this study’s investigation of the impact of hospital
structure on EHR use before (Model 1) and after (Model 2) controlling for hospital
covariates. None of the structural factors was statistically associated with EHR use.
Compared to the hospitals without any contracts and not maintaining a multi-hospital
system, hospitals with the contract and any affiliation had approximately 1 more EHR
functionality, after controlling hospital characteristics.
Table 23. Impact of hospital structure on EHR use Independent variables
Model 1 (Beta¹)
Model 2 (Beta)
Foundation type (private) (ref=public) 0.338 Multi-hospital system (yes) (ref=no) 0.773* Contracts with other hospitals (group purchasing, etc.)(ref=no) 0.893† Location (urban) (ref=rural) 0.235
Structure before EHR adoption
Dec. decision-making 0.011 -0.001 IT staff specialization 0.066 0.034 IT infrastructure 0.153 0.115 Organic structure -0.182 -0.194
* < 0.05, ** < 0.005, ¹ Beta stands for regression coefficient. †: p=0.0502
87
4.3.4. EHR use and environment
This section describes the impact of environmental complexity on EHR use before
(Model 1) and after (Model 2) controlling for hospital covariates.
Hypothesis Addressed:
HU3: Hospitals with high environmental complexity were more likely to use an EHR system.
Table 24 shows that environmental complexity was not related to EHR use as measured
by the number of EHR functionalities. Hospitals affiliated or having contracts had more
EHR functions than their counter parts.
Table 24. Impact of environmental complexity on EHR use Independent variables
Model 1 (Beta¹)
Model 2 (Beta)
Foundation type (private) (ref=public) 0.347 Multi-hospital system (yes) (ref=no) 0.811* Contracts with other hospitals (group purchasing, etc.)(ref=no) 0.939* Location (urban) (ref=rural) 0.239 Environmental complexity (All hosp. within area) -0.007 -0.016
* < 0.05, ** < 0.005, ¹ Beta stands for regression coefficient. Note: the square root of beds used in this model.
4.3.5. EHR use, task, structure, and environment
Table 25 presents the results of analyses investigating whether hospital structural factors
work as moderators between task complexity and EHR use and between environmental
complexity and EHR use.
88
Hypothesis Addressed:
HU4: The relationship between task complexity and the use of EHR would be
dependent on hospital structure.
HU5: The relationship between structure and EHR use be dependent on environmental
complexity.
There were not any interaction effects between task complexity and hospital structural
features or between task complexity and environmental complexity. Affiliation status
and contract status were still significant predictors of EHR use.
Table 25. Impact of a hospital’s internal features on EHR use Independent variables Beta
Intercept 3.644 Foundation type (private) (ref=public) 0.411 Multi-hospital system (yes) (ref=no) 0.829* Contracts with other hospitals (group purchasing, etc.)(ref=no) 0.897† Location (urban) (ref=rural) 0.320 Task complexity # of specialties 0.011
Structure before EHR adoption
Decentralization -0.021 IT staff Specialization -0.006 IT infrastructure -0.203 Organic structure 0.224
Envir. Complexity All hosp. within area -0.025 Task complexity*Structure
Task*Decentralization -0.009 Task*Specialization -0.013 Task*IT infrastructure 0.015 Task*Organic structure 0.055
Environmental complexity*Structure All hosp. within area*Decentralization 0.024 All hosp. within area*Specialization -0.001 All hosp. within area*IT infrastructure 0.029 All hosp. within area*Organic structure 0.032
* < 0.05, ** < 0.005, ¹ Beta stands for regression coefficient. †: p=0.063 Note: all numeric values were analyzed using mean centered predictors. The square root of beds used in this model.
89
4.4. Interoperability
This section reports the investigation of the relationships among hospitals’ internal
features and EHR interoperability using relevant statistical methods described in the
research methods section.
4.4.1. Hospital characteristics by EHR interoperability
Among 158 hospitals adopting an EHR system, thirty-two percent of hospitals (51/158)
reported that they have any kind of EHR interoperability, partially or fully implemented,
on 8 items on patients’ clinical information. Those hospitals reported that they had used
the following tools for patients’ clinical information exchange: e-mail with attached
document, internet access to patient records, patient card system such as IC card, health
information technology system to system, and others.
Table 26. Characteristics of hospitals adopting an EHR system by interoperability
Basic characteristics Hospitals
having EHR interoperability
Hospitals not having EHR
interoperabilityN p-value ¹
% by interoperability status 32.3 67.7 158 (100.0) -
Foundation Private 26.1 73.9 69 (100.0) 0.1427 Public 37.1 62.9 89 (100.0)
Multihospital systems Yes 36.4 63.6 77 (100.0) 0.2842 No 28.4 71.6 81 (100.0)
Contracts for patient delivery
Yes 35.5 64.5 141 (100.0) 0.0128 No 5.9 94.1 17 (100.0)
Location Urban 32.0 68.0 147 (100.0) 0.7475 Rural 36.4 63.6 11 (100.0)
¹ Chi-square test, but Fisher’s Exact Test is applied for contracts for patient delivery and location
Table 26 shows the relationship between EHR interoperability and a hospital’s general
characteristics. While thirty-five percent of hospitals with contracts for patient delivery
and group purchasing reported having EHR interoperability, only 5.9% of hospitals not
having the contract reported having EHR interoperability (p=0.0128).
90
Table 27 presents the results on hospital’s internal features by EHR interoperability
status. The number of medical specialties measuring task complexity was significantly
higher in hospitals with EHR interoperability than hospitals without EHR
interoperability (p=0.0006).
Table 27. Hospital’s internal features by EHR interoperability status Basic characteristics Having
InteroperabilityNot having
Interoperability N (%) p-value²
N 51 107 158 - Task # of medical specialties 15.59 11.94 158 0.0006
Structure
Decision making (average) 3.32 3.14 158 0.4298 Hiring 3.04 2.97 158 0.7818 Promotion 2.90 2.63 158 0.2790 Program 4.10 3.85 158 0.3018 Policy 3.25 3.09 158 0.5116
IT staff specialization¹ 1.62 1.73 158 0.0960 IT infrastructure 7.80 7.90 158 0.8879 Organic structure 2.80 2.71 158 0.6263
Environment HHI score 0.32 0.29 158 0.4581 # of hospitals within the area 8.92 8.65 158 0.7723
¹ Standardized using bed size: the number of IT staff/bed*100 ² p-value of Wilcoxon’s Rank Sum test or chi-square test
4.4.2. EHR interoperability and task complexity
Table 28 presents the impact of task complexity on EHR interoperability before (Model
1) and after (Model 2) controlling hospital characteristics.
Hypothesis Addressed:
HI1: Hospitals with high task complexity were more likely to have interoperability.
There was significant association between task complexity and the odds of having EHR
interoperability. The odds of having EHR interoperability increase by 8.8% for a 1-unit
increase in the number of medical specialties, after controlling for hospital covariates.
91
Table 28. Impact of task complexity on EHR interoperability
Independent variables Model 1 (Odds ratio)
Model 2 (Odds ratio)
Foundation type (private) (ref=public) 1.033 Multi-hospital system (yes) (ref=no) 0.854 Contracts with other hospitals (group purchasing, etc.)(ref=no) 7.223 Location (urban) (ref=rural) 1.210 Task Complexity (# of specialties) 1.092* 1.088*
* < 0.05, ** < 0.005
4.4.3. EHR interoperability and structure
Table 29 presents the impact of hospital structure on EHR interoperability before
(Model 1) and after (Model 2) controlling for a hospital’s general characteristics.
Hypothesis Addressed:
HI2: Hospitals with decentralized decision making, high IT specialization, high IT
infrastructure, and an organic managerial structure were more likely to have
interoperability.
After adjusting for hospital covariates, the odds of reporting EHR interoperability were
estimated to be increased by 48.4% for a 1-unit increase in the score representing
organic structural form though the increase was marginally significant.
Table 29. Impact of hospital structure on EHR interoperability Independent variables Model 1
(Odds ratio) Model 2
(Odds ratio) Foundation type (private) (ref=public) 0.481 Multi-hospital system (yes) (ref=no) 0.879 Contracts with other hospitals (group purchasing, etc.)(ref=no) 10.57* Location (urban) (ref=rural) 1.252
Structure before EHR adoption
Dec. decision-making 1.136 1.098 IT staff specialization 0.968 0.967 IT infrastructure 0.846 0.799 Organic structure 1.295 1.484†
* < 0.05, ** < 0.005, † p=0.0557
92
4.4.4. EHR interoperability and environment
This section investigates the relationship between EHR interoperability and
environmental complexity represented by two variables: existence of true competitors
and the number of hospitals within the local area (county district). Due to the high
correlation issue between HHI and the number of hospitals within a local area, this study
only used the latter variable in the model.
Hypothesis Addressed:
HI3: Hospitals with high environmental complexity were more likely to have
interoperability.
The study result demonstrates that environmental complexity has no significant impact
on EHR interoperability, either before or after controlling for hospital covariates (Table
30).
Table 30. Impact of environmental complexity on EHR interoperability Independent variables Model 1
(Odds ratio) Model 2
(Odds ratio) Foundation type (private) (ref=public) 0.653 Multi-hospital system (yes) (ref=no) 0.891 Contracts with other hospitals (group purchasing, etc.)(ref=no) 9.460* Location (urban) (ref=rural) 1.270 True competitor 0.491 0.603 Environment complexity (All hosp. within area) 1.008 1.002
* < 0.05, ** < 0.005
4.4.5. EHR interoperability, task, structure, and environment
This section reports the investigation of whether or not hospital structural factors work
as moderators between task complexity and EHR interoperability and between
environmental complexity and EHR interoperability.
93
Hypothesis Addressed:
HI4: The relationship between task complexity and interoperability would be
dependent on hospital structure.
HI5: The relationship between structure and interoperability would be dependent on
environmental complexity.
Table 31 presents the results of the regression analysis evaluating outcome variable as
the odds of having EHR interoperability in hospitals. Among the hospital structural
variables, task complexity, IT staff specialization, organic structural factors, and
environmental factors had direct or indirect effects on EHR interoperability.
Table 31. Impact of a hospital’s internal features on EHR interoperability Independent variables Odds ratio
Intercept 0.053 Foundation type (private) (ref=public) 0.959 Multi-hospital system (yes) (ref=no) 0.681 Contracts with other hospitals (group purchasing, etc.)(ref=no) 9.346 Location (urban) (ref=rural) 1.099 Task complexity # of specialties 1.116*
Structure before EHR adoption
Decentralization 0.951 IT Staff Specialization 1.003 IT infrastructure 1.580 Organic structure 1.049*
Envir. Complexity All hosp. within area 1.009 Task complexity*Structure
Task*Decentralization 0.982 Task*Specialization 1.103* Task*IT infrastructure 1.096 Task*Organic structure 0.947
Environmental complexity*Structure All hosp. within area*Decentralization 0.986 All hosp. within area*Specialization 1.037* All hosp. within area*IT infrastructure 1.041 All hosp. within area*Organic structure 1.023
* < 0.05, ** < 0.005 Note: all numeric values were analyzed using mean centered predictors
For the interaction effect between task complexity and IT staff specialization, the impact
of task complexity on the log odds of having EHR interoperability depends on IT staff
94
specialization. The log odds of having EHR interoperability increases as task complexity
increases at a high level of IT staff specialization which is above the average of the
group (Figure 11). However, the log odds of having EHR interoperability decrease when
task complexity increases at a lower level of IT staff specialization which is below the
average of the group (Figure 12).
Figure 11. Task complexity, IT staff specialization, EHR interoperability 1
95
Figure 12. Task complexity, IT staff specialization, EHR interoperability 2
For the interaction effect between environmental complexity and IT staff specialization,
the impact of IT staff specialization measured by IT staff per 100 beds on the log odds of
having EHR interoperability was heavily influenced by environmental complexity.
Assuming that the distinction point of environmental complexity on whether it is high or
low complexity is based on above or below the average number of hospitals within the
local areas, the log odds of having EHR interoperability increases as IT staff
specialization increases at a high level of environmental complexity (Figure 13). In
contrast, the log odds of having EHR interoperability decrease as IT staff specialization
increases at a lower level of environmental complexity (Figure 14).
Figure 13. Environmental complexity, IT staff specialization, EHR interoperability1
96
Figure 14. Environmental complexity, IT staff specialization, EHR interoperability2
97
CHAPTER V DISCUSSION
5.1. Study methods and related issues
This section will review the research methods of this study, focusing on the study design
and study population.
5.1.1. Study design
The purpose of this study is to investigate the impacts of hospitals’ internal features
including task complexity, structure, and environmental factors on hospitals’ EHR
adoption, use, and interoperability. This study design is similar to a case-control study
design comparing two groups in which one has a specific outcome variable and the other
is without that outcome variable. The factors suspected to be related to the outcome
variables are usually compared with each other in a case-control study design.129
Table 32. Comparison of study designs Group Study design in this study Case-Control study design Static-Group Comparison
design
Time Past Current Past Current Past Current
Study O X O† O X X O
Control O O O
†: # of specialty, # of hospitals within local area, hospital’s organic features were measured at the time of survey * O: observation, X: exposure to an event or having an event Campbell and Stanley (1966) proposed 16 types of experimental and quasi-experimental
study designs for researchers.130 Among them, the Static-Group Comparison design is
similar to the design of this study in that it has two groups. However, our study’s
purpose is to measure which factors affected an event which has already happened. Thus,
98
this study slightly modified the Static-Group comparison model (Table 32). There are
several reasons that this study did not exactly follow the case-control or Static-Group
Comparison study designs. First, this study did not know when the hospitals in the study
group adopted an EHR system. Thus, this study compared several factors from the past
of hospitals that currently have an EHR system with the same characteristics of hospitals
that do not currently have an EHR system measured at the time of the survey. Second,
the purpose of this study is not to investigate the impact of an event on the hospital
structure, but to find out which factors affect a certain event. Thus, observation before an
event occurred was necessary. Third, unlike a study on individual patients or animals,
the unit of analysis of this study is hospitals, which are difficult to randomize. Although
these features differ from traditional study design, this study contends that they would be
sufficient to address the hypotheses of this study.
For these reasons, decentralization of decision-making, IT staff specialization, and IT
infrastructure of hospitals before EHR adoption were compared with hospitals that do
not currently have an EHR system. However, task complexity, organic management
structure, and environmental complexity were measured for all hospitals at the time of
survey. Although one may question whether it is reasonable to compare some
hypothesized factors from past observations of hospitals which have adopted an EHR
system with the same factors from current observations of different hospitals, it can shed
some light on possible causality because it does observe hypothesized factors before an
occurrence of EHR adoption. However, this study does make the assumption that these
factors do not change significantly over time and that what characterizes non-adopters
now would also characterize them in earlier time periods when adopters were making
99
their decision. One alternative that could avoid the issue of comparing observations from
different times would be collecting time-stamped data and conducting a survival analysis.
However, this method would be particularly difficult to implement for hospitals who had
not adopted an EHR system. For these reasons, this study asserts that its design provides
reasonable information about the factors that could play a causal role in EHR adoption,
use, and interoperability in hospitals.
5.1.2. Study population
The survey results showed that general hospitals were over-represented in the study
population compared to small hospitals. Approximately 73.0 percent of general hospitals
responded to this study and, in contrast, only 19 percent of small hospitals participated
in this study although small hospitals constitute 73.2 percent of the total hospital
population in Korea. The reason for this difference may be mainly due to lack of IT
departments in small hospitals. Many general hospitals have used various IT
technologies both clinically and administratively, such as electronic billing and
healthcare expense reimbursement. In these cases there was a director or manager in
charge of the daily activities in the IT department. When the primary investigator
requested a hospital’s participation in the survey, they might have been more likely to
participate because they were in charge of IT. In contrast, small hospitals were less
likely to have an IT department and someone who would be interested in responding to
the survey. This would lead to a lower response rate from small hospitals.
Although this study did not consider the existence of an IT department or types of
hospitals (such as general hospital or small hospital) in the study model, this study
100
argues that there are several reasons that these facts did not critically affect the study
results. First, most of the hospitals that participated in the survey were general hospitals.
Thus, current study results may best be generalized to the general hospitals and less so
for small hospitals. Second, inclusion of type of hospital might produce more distorted
study results because the types of hospital are highly correlated with the number of
medical specialties (or task complexity measure) which was already included in this
study. Third, even though some small hospitals did not have IT departments, most small
hospitals had at least one staff member who worked with IT for the purposes of claiming
process or reimbursement management, and they participated in this study.
5.2. Research results
This section will discuss the research results of this study, focusing on EHR adoption,
use, and interoperability.
5.2.1. Hospital’s internal features affecting EHR adoption
General characteristics: The reported EHR adoption rate was 45% of all responding
hospitals. This is slightly higher than Park’s (2005) study result conducted in 2004
which showed that 38% of non-teaching general hospitals had adopted an EHR system.
However, according to Chae’s (2005) study conducted in 2005, the EHR adoption rate
was approximately 20% of hospitals in a nationwide survey.4 Considering a 4-year time
interval between this study and Chae’s (2005) study, the number of Korean hospitals
that have adopted an EHR system seems to have more than doubled from 2005 to 2009.
The difference from Chae also could be due to the over-representation of general
101
hospitals because 73 percent of the general hospitals in Korea responded to this study
and, in contrast, only 19 percent of small hospitals. The size of a hospital generally
affects EHR system adoption.8 In the U.S., depending on the study results, the overall
EHR adoption rate in hospitals is 20-40%7, which is similar to the adoption rate in South
Korea.
Hospitals located in an urban area were more likely to adopt an EHR system. This result
is similar to studies which focus on U.S. hospitals. According to the AHA (2007) study
results, the EHR adoption rate was higher at hospitals located in urban areas than those
located in rural areas.8 Hospitals that had contracts for group purchasing and patient
delivery with other hospitals were more likely to adopt an EHR system than hospitals
without contracts. Necessity of information exchange with other hospitals might
instigate the adoption of an EHR system for those hospitals that had contracts because
EHR systems would provide remote access to clinical information that would facilitate
patient transfers. Hospitals might be influenced to adopt new technologies through the
other contracted networked hospitals, which might affect the adoption of an EHR system
for those hospitals.
Hospital’s internal features: This study proposed that hospitals with high task
complexity, as measured by the number of medical specialties, would be more likely to
adopt an EHR system because high task complexity generates a higher work load that
could be addressed by IT systems, including EHRs, as a way to more cost effectively
handle that workload. There was a demonstrated relationship between EHR adoption
and task complexity which would support this hypothesis.
102
Among a hospital’s four measured structural characteristics, IT infrastructure and
organic management structure were individually associated with EHR adoption.
Hospitals with high IT Infrastructure, as measured by the number of different clinical
information systems, were more likely to be adopters of EHR systems. If hospitals are
already equipped with various information systems, this suggests that these hospitals
maintain a more sophisticated IT infrastructure and have the experience necessary to
take on the implementation of an EHR system. This means that they are more ready for
the adoption of EHR systems. This result is also in accordance with the results of
Anderson (2007) who found a positive relationship between EMR system adoption and
IT infrastructure.
This study also found that hospitals with a management structure that is more organic
were more likely to be adopters of EHR systems. However, it is not clear whether
organic structural factors lead to EHR system adoption or whether adopting an EHR
system causes a hospital’s structure to be more organic since this study measured
organic factors at the time of survey for all hospitals. However this study contends that
the former is more likely to occur because organization culture such as organic structure
is usually built up over a long period of time and may not be changed easily. The
organic structural form has many unique characteristics such as a horizontal rather than a
vertical direction of communication, a network rather than hierarchic structure of control
and authority,131 and mutual agreement among co-workers rather than high reliance on
rules and standard procedures.132 These kinds of features usually provide hospitals with
an environment in which adoption of innovations including EHR systems can occur
more easily. This study result is in accordance with other studies on the relationship
103
organic structural features and organizational innovation described in the literature
review sections such as Aiken,93 Meadows,94 and Hull.95 Judging from these study
results, this study argues that hospital’s organic structural characteristics might critically
affect their EHR system adoption.
In terms of measured environmental factors, EHR adoption was related to the number of
hospitals within a local area. This result is in accordance with other studies on the
relationship between environmental complexity and technology adoption. Abdolrasulnia,
et al (2008) found that medical doctors practicing in counties with high concentration of
physicians were more likely to adopt an EHR system.133 A possible interpretation of this
study result is that competition forces hospitals to adopt EHR systems because hospitals
located in competitive markets might have to use various technologies including EHR
systems in order to increase their competitive power through increased managerial
efficiency or simply to be viewed favorably and have a better image with respect to their
peers.
Interaction effects between task complexity and hospital structure: This study found
that the level of task complexity had an impact on how different components of the
hospital structure were associated with EHR adoption. While task complexity by itself
was positively related to EHR adoption, its association was also moderated by the level
of decentralization of the decision-making system and level of IT infrastructure.
The likelihood of EHR adoption increases as task complexity increases when hospitals
are under a highly decentralized decision-making system. In contrast, under a highly
centralized decision-making system, the likelihood declines as task complexity increases.
From this, we can conclude that a hospital’s decision-making system plays an important
104
role of a moderator between task complexity and an EHR system adoption. At high
levels of decentralization this study is in partial accordance with other research in which
adoption of innovations including new technologies occurs more frequently in
decentralized-decision making system organizations.134,135 However the reverse effect
occurred when there was centralization of the decision-making system. This is contrary
to our initial hypothesis and the explanation for why this might be the case is not evident.
A possible explanation would be that hospitals’ adoption of EHR might be delayed or
lagged because of strong bureaucratic and administrative procedures under the
centralized decision-making system. They might want to see other hospitals adopting an
EHR system before deciding if an EHR system provides them with clinical and
managerial efficiency.
This study also observed that the likelihood of EHR adoption declined as task
complexity increased when hospitals reported high levels of IT infrastructure. In contrast,
the likelihood increased as task complexity increased when hospitals reported a lower
level of IT infrastructure. The latter case can be explained in that hospitals with a lower
level of IT infrastructure may adopt EHR systems in order to increase their capacity for
dealing with a heavy work load coming from task complexity. However, the former case
may need further explanation. Hospitals with high level of IT infrastructure might
already hire enough human resources. Current IT infrastructure could also have enough
capacity to deal with work load coming from task complexity without EHR systems.
Thus, there would be a declining relationship between task complexity and EHR
adoption under a high level of IT infrastructure.
105
Summary: This study found that task complexity, IT infrastructure, organic structure,
and environmental competition were associated with EHR adoption. Since it is unlikely
that EHR adoption drives the number of specialties in a hospital and that the number is
likely to change slowly over time, the most likely explanation is that task complexity is
driving EHR adoption. A similar argument can be applied for organic structure and
environmental complexity, which are characteristics that are not likely to change within
a short period of time. Thus, this study contends that they would be more likely to affect
EHR system adoption rather than the reverse although they were measured at the time of
survey. Past experience and hospital culture emphasizing flexibility among co-workers
might instigate EHR adoption in hospitals with high levels of IT infrastructure and
organic management structures. Motivation due to increasing competitive pressure from
surrounding peer hospitals might also influence EHR system adoption. With respect to
relationship between task complexity and EHR adoption, bureaucratic procedures under
high centralization of decision-making and potential slack capacity of IT infrastructure
in hospitals with high levels of IT infrastructure appear to differentially affect EHR
adoption.
5.2.2. Hospital’s internal features affecting EHR use
General characteristics: A multi-hospital system measured by whether a hospital is
legally affiliated with other hospitals or not and hospital’s contract with other hospitals
for patient delivery and group purchasing exhibited a greater degree EHR use as
measured by the number of EHR functionalities. Multi-hospital systems might have
more resources to invest in sophisticated IT, including EHRs, than single hospitals,
106
which might result in having more EHR functions. However, it is not clear why
hospitals with the contracts exhibited higher levels of EHR use, and further research is
needed to investigate this issue.
Hospital’s internal features: None of the hospital’s internal features were related to
EHR use. This result suggests several implications for future studies.
First the measure of EHR use might not be properly measuring what this study intended
it to measure. The existence of certain subsystems may not be reflective of overall use of
the system. Use can be defined as the degree to which an EHR is used in patient care,
where the maximum level is when an EHR is used for every patient in every care
process. What this study is measuring may be the expanded potential uses of an EHR
and not its actual use. Rather than simply counting the number of functions in an EHR
system, EHR use may need to be measured more quantitatively. Such measures might be
the number of times data retrieval occurred per month, the number of medical
professions that used an EHR system, and how many logged on hours an EHR system
exhibited. Theoretically, EHR use has a high possibility of correlation with patient care
factors such as frequent patient visits that would increase the retrieval of patient charts
under the paper-based medical record systems. Similarly, increased information
processing demands due to increased admissions or diversity of patients may lead to
increased EHR use. EHR use should be an actual use of EHR system and one wholly
replacing the paper-based medical record system and reflecting patient treatment process
rather than counting the number of functions in an EHR system, which would be a better
measurement and provide more meaningful information for users such as hospital
107
managers, health informaticians, and policy makers. However this information is
difficult and expensive to collect and has not been done in other studies to date.
Alternatively, the finding that none of the hypothesized internal structural factors were
related to EHR use measured by EHR functions may be correct. In this case it might be
better to measure performance factors or increasing information processing demands
such as patient admissions and bed turnover rates. A recent study reflects this thinking.
It investigated the relationship between information processing demand and HIT use in
hospitals and found that increased information processing demand, measured by an
increase in annual patient admissions, was positively associated with HIT use.136 It
measured HIT use by counting different IT systems and by weighting EHR. A second
study looked at the relationship between computerization and quality of care. They
found that there was a positive relationship between the two factors.137 Thus,
performance factors such as reducing or increasing information demand and improving
quality of care may be more reasonable predictors of EHR use.
Lastly, there have been few previous studies on EHR use as an outcome variable at the
organizational level. Most current studies measure EHR use as existence of various
functionalities in an EHR system. They do not use those measures as one of the outcome
variables. Certainly, it is necessary to develop a more quantitative measure of EHR use
mentioned above and to compose hypotheses on how a hospital’s task and information
process demand are related to EHR use, which will help us to understand the mechanism
of variation of actual EHR use.
Summary: Given that EHR use measured by the number of functions in EHR systems
was not associated with any of hospital’s structural factors, alternative measures of EHR
108
use and internal features selected in different empirical and theoretical standpoints were
suggested.
5.2.3. Hospital’s internal features affecting EHR interoperability
General characteristics: A hospital characteristic, contract status with other hospitals
for group purchasing and patient delivery, was associated with hospitals’ EHR
interoperability. Having a relationship with other hospitals that involves movement of
patients from one to another might explain an increased level of interoperability to
exchange clinical information with other hospitals and could explain this finding.
Hospital’s internal features: Task complexity was associated with EHR
interoperability as measured by the ability of an EHR system to exchange patient’s
clinical information electronically with other healthcare organizations. High task
complexity means that hospitals provide diverse patients with comprehensive medical
care services from primary care to specialty care. They would have many cases being
transferred to or being transferred from other clinics or hospitals. All these cases would
increase the necessity of healthcare information exchange and EHR interoperability.
This could be a possible explanation of this finding.
The degree of organic management structure as measured by questions on a hospital’s
organic characteristics was marginally associated with the level of EHR interoperability.
Such a structure has characteristics that could lead to greater interoperability. As Child
described, organic structural form incorporates many flexible features such as lateral
communications network, decentralized decision making, and a high reliance on
consensus among co-workers.132 Active collaborations or cooperative, network-based
109
communications among member or networked hospitals may encourage the idea of EHR
interoperability. Employees working at those hospitals may place more emphasis on the
necessity of patient information. The study result is also in accordance with the other
study results that organizational innovation more frequently occurs within organizations
with more organic than mechanical managerial structures.94, 95
Interaction effects between task complexity and hospital structure: This study found
one structural factor that played an important moderating role between task complexity
and EHR interoperability and between environmental complexity and EHR
interoperability: IT staff specialization as measured by the number of IT staff positions
per 100 beds.
The likelihood of EHR interoperability at a high level of IT staff specialization gradually
increased as task complexity increased as we hypothesized. At low levels of IT staff
specialization the likelihood slightly decreased as task complexity increased. This latter
result contradicts our hypothesis that task complexity would increase EHR
interoperability. It may be that a competition for resources under high task complexity
might cause scarce resources to use for specialty care needs such as hiring more nurses
or purchasing medical equipment rather than hiring IT staff for developing
interoperability in an EHR system.
This study also found that the likelihood of EHR interoperability at a high level of
environmental complexity increased with increasing IT staff specialization. However,
the likelihood of EHR interoperability at a lower level of environmental complexity
decreased with increasing IT staff specialization. Possible interpretation of the latter
result would be that hospitals in a less complex environment might have less motivation
110
to implement EHR interoperability although they had enough IT staff specialization
because there are few competitors. Due to lack of similar studies, this study suggests that
it is necessary to do further research.
Summary: In summary, this study speculates that high task complexity requiring a
comprehensive medical care through patient’s referral would lead to increase the
likelihood of having EHR interoperability. The study result that hospitals with an
organic managerial structure had a higher likelihood of EHR interoperability confirmed
other findings in the literature. EHR interoperability can be seen as an innovative event.
This study speculates that high competition among medical specialties might force
hospitals’ resources toward specialty care areas at low levels of IT staff specialization,
which might lead to decreased EHR interoperability under high task complexity and a
lower level of IT staff specialization. A hospital’s weak motivation under a lower level
of environmental complexity might lead to decrease the likelihood of EHR
interoperability even though there was a high level of IT staff specialization.
5.3. Study limitations
Specific issues regarding research methods and results were partially discussed above. In
addition, this study has the following limitations. One involves the generalizability of
the study results. This study obtained the data from South Korean hospitals. A study
using data from a different country could produce different study results. Thus, the
results interpretation may be limited to hospitals in Korea. It may be necessary to further
study the proposed hypotheses using data from hospitals in other countries such as the
United States.
111
Another limitation may be a selection bias in study subjects. This study advertised to
hospitals’ CIOs or IT managers one time through postal mail. In addition, some hospitals
were called using a telephone number from the Korea Hospital Association in order to
increase the response rate. Due to the limitation of the data collection period, only two
hospitals in each local area were selected based on the Korean language’s alphabetic
order of hospital address. This study also received data collection support from a
professional community of hospital IT department managers in Korea. Most association
members responded to the survey. One significant characteristic of this association is
that most of the members are IT managers of general hospitals, which are larger sized
hospitals. This was reflected in the over-representation of general hospitals in the survey
responses. Thus there might be some selection bias in which the study results may
represent larger Korean hospitals or general hospitals with IT departments.
An informant recollection issue might have affected the quality of survey data. A
number of the survey questions were based on recollection of past events and
organizational characteristics, such as the year of EHR installation. To respond
accurately the informant needed to know the past history of the hospital, at least for the
past 6 years. Depending on the informant, a six-year recollection period may be too long
a period for accurate recall. However, it is likely that the installation of an EHR system
would be a significant event for hospitals, so this study argues that IT managers should
be able to remember its introduction year. Thus possibility of incorrect responses in this
survey would be reduced.
112
CHAPTER VI CONCLUSIONS
6.1. Study overview
The objective of this study was to investigate the relationships between EHR adoption,
use, and interoperability and hospitals’ various contexts such as task complexity,
hospital structure, and environment. For this purpose, this study proposed that EHR
adoption, use, and interoperability would be positively related to 1) task complexity, 2)
structural characteristics such as decentralization of decision-making, IT staff
specialization, IT infrastructure, and a hospital’s organic structural characteristics, and 3)
environmental complexity. This study also proposed that the impacts of task complexity
and environmental complexity on EHR adoption, use, and interoperability would depend
on hospital’s structural characteristics.
This study descriptively defined EHR adoption, use, and interoperability and conducted
a nationwide EHR survey of the IT departments in Korean hospitals. The survey was
implemented from April 1 to August 3, 2009 through a University of Minnesota internet
website. Target survey informants were the chief information officers (CIO) and IT
department managers in hospitals in South Korea. Among a hospital population of 1,067,
a total of 356 hospitals participated in the survey. The final response rate was 33.5% of
the total hospital population.
In the stage of statistical data analysis, this study used Generalized Estimating Equations
(GEE), an extension of the Generalized Linear Model, in the case of EHR adoption and
113
interoperability and General Linear Mixed Model with random intercept statement in the
case of EHR use.
6.2. Significant findings
The study reveals that the hospital’s internal structural features were related to EHR
system adoption and EHR interoperability. However, they were not associated with EHR
use. The following is a brief summary of significant findings in this study by three parts
–one for each outcome.
6.2.1. Findings in EHR adoption
The likelihood of EHR adoption was related to a hospital’s internal features and
environmental factors. Figure 15 depicts and summarizes the impact of hospital structure
and environmental factors on EHR adoption verified by this study.
First, HA1 was supported. The likelihood of EHR adoption increased as hospital’s task
complexity measured by the number of medical specialties increased.
Second, HA2 was supported by IT infrastructure and organic management structure. The
likelihood of EHR adoption increased as hospital’s IT infrastructure and organic
structural characteristics increased. However, EHR adoption was not individually related
to decentralization of decision-making system and IT staff specialization.
Third, HA3 was supported. The likelihood of EHR adoption increased as the number of
hospitals within a local area increased.
114
Fourth, HA4 was supported by two areas. The impact of task complexity on an EHR
system adoption was dependent on decentralization of decision-making and IT
infrastructure, respectively. Assuming that a hospital adds additional medical specialties
or a hospital’s task complexity increases, the likelihood of adopting an EHR system of
the hospital increases under a decentralized decision-making system, but decreases
under a centralized decision-making system. Under the same condition of task
complexity, the likelihood of adopting an EHR system of the hospital decreases at a high
level of IT infrastructure, but increases at a lower level of IT infrastructure.
Fifth, HA5 was rejected. There was no combined effect of environmental complexity
and hospital structure on EHR adoption.
115
Figure 15. Impacts of hospital structure & environment on EHR adoption
6.2.2. Findings in EHR use
Unlike our expectations, neither a hospital’s internal features nor its environmental
complexity nor their combination were related to the degree of EHR use. Thus,
hypotheses from HU1 to HU5 were rejected. Alternative measures of EHR use and a
hospital’s internal features were suggested through these findings.
6.2.3. Findings in EHR interoperability
The likelihood of EHR interoperability was related to a hospital’s internal features and
environmental factors. Figure 16 depicts and summarizes hospital structure and
environmental factors affecting EHR interoperability observed by this study.
First, HI1 was supported. The likelihood of EHR interoperability increased as a
hospital’s task complexity measured by the medical specialty increased.
Second, HI2 was supported with respect to organic management structure though its
statistical significance was marginal. Hospitals having greater organic structural
characteristics were more likely to have an interoperable EHR system. However,
decentralization of decision-making system, IT staff specialization, and IT infrastructure
was not individually associated with an interoperable EHR system.
Third, HI3 was rejected. Environmental complexity measured by the number of
hospitals within the local area was not related to EHR interoperability.
116
Fourth, HI4 was supported by the combined effect of task complexity with IT staff
specialization on EHR interoperability. Assuming that a hospital adds additional medical
specialties, the likelihood of EHR interoperability of the hospital increases at a high
level of IT staff specialization, but decreases at a lower level of IT staff specialization.
Other than that, none of combined effect of task complexity with a hospital’s internal
features was reported.
Figure 16. Impacts of hospital structure & environment on EHR interoperability
Fifth, HI5 was supported by the combined effect of environmental complexity with IT
staff specialization on EHR interoperability. At a high level of environmental
complexity where the number of hospitals within the local area is above the average of
117
the study population, the likelihood of having EHR interoperability increases as IT staff
specialization increases. However, the likelihood of having EHR interoperability
decreases as IT staff specialization increases at a lower level of environmental
complexity where the number of hospitals within the local area is below the average of
the study population. Other than that, none of the combined effect of environmental
complexity with structural features on EHR interoperability was reported.
6.3. Contributions to the field of health informatics
EHR systems have been rapidly adopted and used in healthcare settings for a relatively
short period of time, and they have recently received a nationwide spotlight on EHR
adoption, use, and interoperability. This study would be a first study on how various
factors surrounding hospitals affect EHR adoption, use, and interoperability. IT
managers working in hospitals and students and scholars in academic settings would
benefit from this new knowledge about how an overall hospital’s structural settings
affect EHR system adoption, use, and interoperability.
Another potential contribution would be one of theory application studies using an EHR
subject. By introducing an academically well-known organizational theory into the field
of informatics, students and informaticians working in empirical fields would get some
experience of theory application. According to a recent comprehensive review study,138
contingency theory has not been used and applied much in the field of health informatics.
Researchers and scholars working in the field of health informatics would get an
opportunity to see how organizational theory helps to predict EHR adoption, use, and
118
interoperability. Studies based on strong theoretical arguments give us insight on a better
logical understanding and prediction of a central phenomenon. This study is one of new
approaches to explaining the impact of technology, structure, and environment on EHR
system adoption, use, and interoperability based on theoretical arguments.
By introducing the field of health informatics from a foreign country to health
informaticians, this study allows them to have a broad viewpoint and picture of the
world concerning task and hospital structure. If we study only IT in specific areas and
discard the opportunity to obtain knowledge from other countries, then we will have a
narrow-minded perspective.
Some research methods and several constructs which were not fully developed in this
study can also be used as an arena of criticism through which health informaticians can
improve their thoughts and reasoning concerning task complexity and organization
structure.
6.4. Future study directions
The following are suggestions for future studies. First, it is definitely necessary for
investigators to develop a measure of EHR use that is more quantitatively based rather
than measured by counting functionalities in an EHR system. The number of retrievals
or total logged-on time per month per medical professional in an EHR system may be
some alternatives. This measure should be focused at the organizational level, rather
than at individual levels because there have been a paucity of studies focusing on the
organizational level.
119
Second, it is also necessary to choose dynamic categories of predictors such as the
number of patient admissions per month, existence of quality assurance, hospital length
of stays, and bed turnover rate in terms of EHR use and interoperability because they
may be closely related with EHR use judging from the fact that frequent patient’s
admissions increase paper-chart delivery to the clinical departments. EHR use and
interoperability also would be affected by those predictors because an EHR system is
quickly replacing a paper-based medical record system.
Thirdly, collecting data that have more hierarchical structures such as individual medical
professionals, hospital structures, and environmental factors can be another important
study direction. By combining different levels of data structure, we can look at how
EHR adoption, use, and interoperability are different or are affected by individual,
organizational, and environmental factors.
Fourth, many scholars have argued that hospitals should introduce more information
technology to reduce medical errors and to improve the quality of care.139, 140 From this
standpoint, further study is needed to investigate the impact of EHR system adoption,
use, and interoperability on clinical outcomes and organizational performance or
managerial efficiency.
In conclusion, this study investigated the impact of the hospitals’ technology, structure,
and environment on EHR adoption, use, and interoperability. By introducing two new
concepts, EHR use and interoperability in addition to EHR adoption, this study
attempted to see their relationship with other hospital predictors that had not been
previously studied. This study anticipates that its results should provide hospital IT
120
managers, health informaticians, and policymakers with new information on the factors
related to EHR adoption, use, and interoperability.
121
* Bibliography
1. Kuhn TS. The structure of scientific revolutions (third Edition). the University of Chicago Press: Chicago and London. 1996.
2. Kim JY, Chae YM. Factors affecting the users’ satisfaction on the electronic medical record system. J Korean Soc Med Inform 2002;8(4):39-53.
3. Park RW, Shin SS, Choi YI, Ahn JO, Hwang SC. Computerized physician order entry and electronic medical record systems in Korean teaching and general hospitals: results of a 2004 survey. Journal of the American Medical Informatics Association. 2005;12(6):642-647
4. Chae YM, Park KM, Kim JE, Kim YE, Lee BH, Kang SW. Status of computerization in Korean health institutions. Health Insurance Review Agency. 2005.12
5. Yu S, Yoo S, Min W. A survey on current status of computerization of physician office practice in Korea. Journal of Korean Society of Medical Informatics. 2003;9(3):323~330
6. Hwang JI, Yu SJ, Chin HJ, Seo JW. Needs assessment for functionalities in electronic health record systems in general hospitals J Kor Soc Med Informatics. 2005;12(1):57-70.
7. Jha AK, Ferris TG, Donelan K, DesRoches C, Shields A, Rosenbaum S, Blumenthal D. How common are electronic health records in the United States? a summary of the evidence. Health Affairs. 2006;Web Exclusives 25:pw496-w507
8. American Hospital Association. Continued progress: hospital use of information technology. Chicago (IL). 2007
9. Felt-Lisk S. New hospital information technology: is it helping to improve quality? issue brief of timely information from Mathematica (Mathematica Policy Research, Inc.), 2006 May;3:1-4
10. Gans D, Kralewski J, Hammons T, Dowd B. Medical groups’ adoption of electronic health records and information systems, Health Affairs. 2005;24(5):1323-1333
11. eHealth Initiative: Foundation for eHealth Initiative. eHealth initiative releases results of 2007 survey on health information exchange, December 19, 2007
12. Adler-Milstein J, McAfee AP, Bates DW, Jha AK. The state of regional health information organizations: current activities and financing, Health Affairs-Web Exclusive, 2007 December 11;w60-w69.
13. Lee CH, Sung YH, Jung YY, Lee JL. A study on the effects of EMR on nursing documentation, Journal of Korean Society of Medical Informatics. 2000;6(4):87~97
14. Noh YA, Choi I, Jeong SO, Kim S. The study on difference in volume of the data between electronic medical record and paper-based medical record: comparison of chief complaint and present illness on medical record, Journal of the Korean Society of Health Informatics and Health Statistics. 2007;32(2):27-37
122
15. Shin AM, Park HJ, Jung SJ, Noh MJ, Kim YN. The paper-based medical record compared to the electronic medical record: documentation and agreement of information, Journal of Korean Society of Medical Informatics. 2009;15(2):201-208
16. Pizziferri L, Kittler AF, Volk LA, Honour MM, Gupta S, Wang S, et al. Primary care physician time utilization before and after implementation of an electronic health record: a time-motion study. J Biomed Inform. 2005;38:176-88.
17. Wong DH, Gallegos Y, Weinger MB, Clack S, Slagle J, Anderson CT. Changes in intensive care unit nurse task activity after installation of a third generation intensive care unit information system. Crit Care Med. 2003;31:2488-94.
18. Hayrinena K, Saranto K, Nykanen P. Definition, structure, content, use and impacts of electronic health records: A review of the research literature. International Journal of Medical Informatics. 2008;77:291-304
19. Shekelle PG, Morton SC, Keeler EB. Costs and benefits of health information technology. Evid Rep Technol Assess. 2006 April;132:1-71
20. Loomis GA, Ries JS, Saywell RM, Jr, Thakker NR. If electronic medical records are so great, why aren't family physicians using them? The journal of Family Practice. 2002 July;51(7):636-641
21. McDonald CJ. The barriers to electronic medical record systems and how to overcome them, Journal of the American Medical Informatics Association. 1997 May;4(3):213-221
22. Payne TH, Perkins M, Kalus R, Reilly D. The transition to electronic documentation on a teaching hospital medical service. AMIA Annu Symp Proc. 2006;629-633.
23. Kim GH. Study on the Korean model for community-based EHR(Electronic Health Record) system, Graduate School of Public Health, Yonsei University, 2005, June.
24. The Center for Interoperable EHR, General overview. 2010. 3. http://ehrkorea.org:8082/about/greeting.aspx (Accessed on April 22, 2010). South Korea.
25. The CMS Office of Public Affairs. CMS proposes definition of meaningful use of certified electronic health records (EHR) technology. 2009 December 30. http://healthit.hhs.gov/portal/server.pt?open=512&objID=1325&parentname=CommunityPage&parentid=1&mode=2
26. Wilson J F. Making electronic health records meaningful. Annals of Internal Medicine. 2009 August;151(4):293-296
27. Massaro TA. Introducing physician order entry at a major academic medical center: I. impact on organizational culture and behavior, Academic Medicine. 1993 Jan;68(1):20-25
28. Kim MY, Kim KH, Jang HS, Yu HS, Seo SW. A study on the medical record technicians manpower relation by before and after computerization of medical record, Journal of Korean Society of Medical Informatics, 1998;4(2):25-34
123
29. Hwang JI, Park HA. Nurses’ experience: impact of hospital information systems on nursing care, J Kor Soc Med Informatics 2005;12(1):83-94
30. Seo J, Kim KH, Choi JW, Ha K, Chin HJ, Kim JU, Kim SW, Im J, Kim S. Implementation of electronic medical records at Seoul National University Hospital, Journal of Korean Society of Medical Informatics, 2006;12(3):213-225
31. Choi EY, Chang Y, Chin HJ, Chung EJ, Byeun NS, Yoo D, Bae S, Ha KS. User’s satisfaction on the electronic medical record system in Seoul National University Bundang Hospital. Journal of Korean Society of Medical Informatics, 2005;12(1):45-55
32. Cherry B. Determining factors of organizational readiness for technology adoption in long-term care facilities, Texas Tech University Health Science Center. 2008 September 28:1-38
33. Park JS, Kim HS, Park YT, Speedie SM. Construct comparisons of IT adoption theories across cultures and innovativeness of health care professionals, a poster summary submitted to AMIA 2008 Annual Symposium, 2008, 11.
34. Wikipedia Foundation Inc., South Korea, http://en.wikipedia.org/wiki/South_Korea (Accessed on December 10, 2009)
35. U.S. Department of State. Background Note: South Korea; 2009 October 30. (Available on the internet at http://www.state.gov/r/pa/ei/bgn/2800.htm) (Accessed on December 10, 2009)
36. Wikipedia Foundation Inc ., Minnesota. http://en.wikipedia.org/wiki/Minnesota. (Accessed on December 11, 2009)
37 .World Health Organization: Regional Office for the Western Pacific. Country context: Republic of Korea. Available on the internet at http://www.wpro.who.int/countries/2009/kor/ (Accessed on December 10, 2009)
38. U.S. Census Bureau. Population estimates and projections. Available at http://www.census.gov/ (Accessed on December 11, 2009)
39. Wikipedia Foundation Inc ., Provinces of Korea. (Accessed on December 10, 2009). http://en.wikipedia.org/wiki/Provinces_of_Korea
40. Lee EJ, Kim SY, Chae YM. Legislation direction for health information privacy in the telemedicine era. J Kor Soc Med Informatics 2009;15(4):361-371
41. The National Assembly of the Republic of Korea. Korean medical law, section 23. Available at http://www.assembly.go.kr/renew09/info/inf/tra_list.jsp?M_idx=2_01 (Access on December 11, 2009), South Korea
42. Ministry of Health and Welfare (MOHW), MOHW News, 12/17/2007
43. American Hospital Association. 2006 AHA survey on hospital adoption of health information technology, November 10, 2006
44. DesRoches CM, Campbell EG, Rao SR, Donelan K, Ferris TG, Jha A, Kaushal R, Levy DE, Rosenbaum S, Shields AE, Blumenthal D. Electronic health records in
124
ambulatory care - a national survey of physicians. The New England Journal of Medicine, 2008 July 3;359(1):50-60
45. Jha AK, DesRoches CM, Campbell EG, Donelan K, Rao SR, Ferris TG, Shields A, Rosenbaum S, Blumenthal D. Use of electronic health records in U.S. hospitals, NEJM. 2009 April 16;360(16):1628-1638
46. Medical Group Management Association Center for Research. Assessing adoption of effective health information technology questionnaire. 2005 January 31
47. Tang PC, Coye MJ, Bakken S, et al. Key capabilities of an electronic health record system. Institute of Medicine: Committee on Data Standards for Patient Safety· Board on Health Care Services, Washington, D.C., 2003
48. Federal Register / Vol. 75, No. 8 / Wednesday, January 13, 2010 / Rules and Regulations
49. Healthcare Information and Management Systems Society. Interoperability definition and background. June 9, 2005. Web site at: http://www.himss.org/ASP/ContentRedirector.asp?ContentID=65063
50. U.S. Government Accountability Office. Information technology - challenges remain for VA’s sharing of electronic health records with DOD. Testimony Before the Subcommittee on Military Construction, Veterans’ Affairs, and Related Agencies; House Committee on Appropriations. GAO-09-427T. March 12, 2009
51. Bouhaddou O, Warnekar P, Parrish F, Do N, Mandel J, Kilbourne J, Lincoln MJ. Exchange of computable patient data between the Department of Veterans Affairs and the Department of Defense: Terminology Mediation Strategy. J Am Med Inform Assoc. 2008;15:174-183.
52. Braithwaite WR, Hammond WE, Van Campen M, Ryan S, Alschuler L, Blocker A, Brandman D. HL7 NLM interoperability survey. The Department of Health and Human Services, HL7, National Library of Medicine (NLM) July 18, 2005.
53. Lawrence PR, Lorsch JW. Organization and environment: managing differentiation and integration, Harvard University Press, 1967
54. Galbraith JR. Designing complex organizations. Reading, MA: Addison-Wesley, 1973
55. Scott WR. Organizations: rational, natural, and open systems. Prentice Hall, Fifth Edition, 2003
56. Hickson DJ, Hinings CR, Lee CA, Schneck RE, Pennings JM. A strategic contingencies theory of intraorganizational power. Administrative Science Quarterly. 1971;16(2):216-229
57. Knoke D. Political Networks: the structural perspective, Cambridge University Press, 1990
58. Orlikowski WJ, Barley SR. Technology and institutions: what can research on information technology and research on organization learn from each other? MIS Quarterly. 2001 June;25(2):145-165
125
59. Flood AB, Scott WR. Hospital structure and performance. The Johns Hopkins University Press. 1987
60. Zinn JS, Mor V. Organizational structure and the delivery of primary care to older Americans. Health Services Research. 1998 June;33(2):354-380
61. Harvey E. Technology and the structure of organizations, American Sociological Review. 1968 April;33(2):247-259
62. Wright PM, Noe RA. Management of organizations, Richard D.Irwin, a Times Mirror Higher Education Group, Inc. Company, 1996
63. Hage J, Dewar R. Elite values versus organizational structure in predicting innovation. Administrative Science Quarterly. 1973;18(3):279-290
64. Hage J, Aiken M. Routine technology, social structure, and organizational goals, Administrative Science Quarterly, 1969;14(3):366-376
65. Daft RL. Organization theory and design. West Publishing Company-Saint Paul, New York, Los Angeles, San Francisco, Fourth Edition, 1992
66. Thompson VA. Hierarchy, specialization, and organizational conflict. Administrative Science Quarterly. 1961 March;5(4):485-521
67. Pfeffer J, Leblebici H. Information technology and organizational structure. the Pacific Sociological Review. 1977 April;20(2):241-26
68. Fombrun CJ. Structural dynamics within and between organizations, Administrative Science Quarterly. 1986 Sep.;31(3):403-421
69. Burns T, Stalker GM. The management of innovation, London: Tavistock, 1961
70. Tung RL. Dimensions of organizational environments: an exploratory study of their impact on organization structure. Academy of Management Journal. 1979; 22(4):672-693
71. Clark T, Varadarajan PR, Pride WM. Environmental management: the construct and research propositions, Journal of Business Research, 1994; 29(1):23-38
72. Duncan RB. Characteristics of organizational environment and perceived environmental uncertainty, 1972 September;17(3):313-327
73. Gebauer J, Shaw MJ, Gribbins ML. Task-technology fit for mobile information systems. Published 6/26/2006 (http://www.business.uiuc.edu/Working_Papers/papers/06-0107.pdf)(accessed May 14, 2010):1-36
74. Goodhue DL. Understanding user evaluations of information systems, Management Science. 1995 Dec.;41(12):1827-1844
75. Goodhue DL, Thompson RL. Task-technology fit and individual performance. MIS Quarterly. 1995;19(2):213-236
76. Ammenwerth E, Iller C, Mahler C. IT-adoption and the interaction of task, technology and individuals: a fit framework and a case study, BMC Medical
126
Informatics and Decision Making 2006, 6(3), page number is not available (http://www.biomedcentral.com/content/pdf/1472-6947-6-3.pdf)
77. Burton LC, Anderson Gerard F, Kues IW. Using electronic health records to help coordinate care. the Milbank Quarterly. 2004; 82(3):457-481
78. Schoenman JA. Small, stand-alone, and struggling: the adoption of health information technology by rural hospitals. Policy Analysis Brief, 2007; 10 National Opinion Research Center (Walsh Center for Rural Health Analysis):1-4
79. McCullough JS. The adoption and diffusion of hospital information systems, The University of Pennsylvania, Ph.D. dissertation, 2005
80. Simon SR, Kaushal R, Cleary PD, Jenter CA, Volk LA, Poon EG, Orav EJ, Lo HG, Williams DH, Bates DW. Correlates of electronic health record adoption in office practices: a statewide survey. Journal of the American Medical Informatics Association. 2007;14(1):110-117
81. Healthcare Financial Management Association. Overcoming barriers to electronic health record adoption, February 2006
82. Miller RH, Sim I. Physicians’ use of electronic medical records: barriers and solutions. Health Affairs. 2004;23(2):116-126
83. The Centers for Medicare and Medicaid Services (CMS), the Substance Abuse and Mental Health Services Administration (SAMHSA). Increasing interoperability in health information systems for Medicaid, Mental Health, and Substance Abuse Treatment, Conference in Washington D.C., ON January 24-25, 2007.
84. Coleman EA, May K, Bennett RE, Dorr D, Harvell J. Report on health information exchange in post-acute and long-term care. University of Colorado at Denver and Health Science Center, Oregon Health and Science University, DHHS (The report was prepared for U.S. Department of Health and Human Services (DHHS): Office of the Assistant Secretary for Planning and Evaluation Office of Disability, Office of Disability, Aging and Long-Term Care Policy, February 2007)
85. Brailer DJ. From Santa Barbara to Washington: a person’s and a nation’s journey toward portable health information. Health Affairs. 2007;26(5):w581-w588
86. Miller RH, Miller BS. The Santa Barbara County care data exchange: what happened?. Health Affairs. 2007; 26(5):w568-w580
87. Healthcare Information and Management Systems Society, 14th annual HIMSS leadership survey: Healthcare Chief Information Officer Results. 2003; February 10, 1-32
88. Jeyaraj A, Rottman JW, Lacity MC. A review of the predictors, linkages, and biases in IT innovation adoption research. Journal of Information Technology. 2006;21:1-23
89. Wang B, Wan TTH, Burke DE, Bazzoli GD, Lin BYJ. Factors influencing health information system adoption in American hospitals. Health Care Manage Rev. 2005 January/March;30(1):44-51
127
90. Moch MK, Morse EV. Size, centralization and organizational adoption of innovations. American Sociological Review. 1977 Oct.;42(5):716-725
91. Chau PYK., Tam KY. Factor affecting the adoption of open systems: an exploratory study. MIS Quarterly. 1997 Mar.;21(1):1-24
92. Anderson C, Mishra AN, Agarwal R, Angst CM, and Labkoff S. The missing link: electronic health record adoption by physician practices. CHIDS Research Briefings (2:1B), Center for Health Information and Decision Systems, Robert H. Smith School of Business, University of Maryland, Spring 2007, p.1-4
93. Aiken M, Hage J. The organic organization and innovation. Sociology. 1971;5(1):63-82
94. Meadows IG. Organic structure and innovation in small work groups. Human Relations. 1980;33(6):369-382
95. Hull F, Hage J. Organizing for innovation: beyond Burns and Stalker's organic type. Sociology. 1982;16(4):564-577
96. Burke DE, Wang BBL, Wan TTH, Diana ML. Exploring hospital’s adoption of information technology,=. Journal of Medical Systems. 2002 August;26(4):349-355
97. Baker L, Wheeler S. Managed care and technology diffusion: the case MRI. Health Affairs 1998; 17(5):195-203
98. Friedman LH, Goes JB. The timing of medical technology acquisition: strategic decision making in turbulent environments. Journal of Healthcare Management. 2000;45(5):317-330
99. Brynjolfsson E, Hitt LM. Beyond computation: information technology, organizational transformation and business performance, the Journal of Economic Perspectives. 2000;14(4):23-48, p.35
100. Pfeffer J, Leblebici H. The effect of competition on some dimensions of organizational structure. Social Forces, 1973;52(2):268-279
101. Kazley AS, Ozean YA. Organizational and environmental determinants of hospital EMR adoption: a national survey. Journal of Medical System. 2007;31:375-384.
102. Barley SR. The alignment of technology and structure through roles and networks. Administrative Science Quarterly: Special issue: technology, organizations, and innovation. 1990 Mar.;35(1):61-103
103. Korean Hospital Association. Hospital roster. 2008. 1.
104. The Office of the National Coordinator for Health Information Technology (ONC), ALLIANCE (The National Alliance for Health Information Technology), the National alliance for health information technology report to the Office of the National Coordinator for Health Information Technology on defining key health information technology terms, April 28, 2008.
105. Garets D, Davis M. Electronic medical records vs. electronic health records: yes, there is a difference, A HIMSS Analytics White Paper (2006 HIMSS Analytics), Updated January 26, 2006
128
106. Texas Medial Association (TMA), Electronic Medical Record vs. Electronic Health Record, Last Published: 6/19/2006, web site at http://www.texmed.org/Template.aspx?id=5072 (visited on July 6, 2008).
107. Bryant RE, O'Hallaron DR. Computer systems: a programmer's perspective, Prentice Hall, 2003
108. Health Insurance Review & Assessment Service. 2005 Hospital information technology survey, 2005, 11.
109. Blumenthal D, DesRoches C, Donelan K, Ferris T, Jha A, Kaushal R, Rao S, Rosenbaum S. Health information technology in the United States: the information base for progress, the George Washington University, Massachusetts General Hospital, Robert Wood Johnson Foundation, 2006
110. Mintzberg H. The structuring of organizations. Englewood cliffs, N.J. Prentice-Hall, 1979
111. Covin JG, Slevin DP. The influence of organization structure on the utility of an entrepreneurial top management style. Journal of Management Studies. 1988. 25, pp. 217-234.
112. Kwon ST. A study on the effects of strategy, competencies, organizational structure and their interactions on the performance of hospitals, Department of Public Health, the Graduate School of Yonsei University, Ph.D. dissertation, 2003.
113. Cannon AR, John CH. Measuring environmental complexity: a theoretical and empirical assessment, Organizational Research Methods, 2007;10(2);296-321
114. Wikipedia Foundation Inc., Herfindahl index. 2010. (at http://en.wikipedia.org/wiki/Herfindahl_index)(Access 14 May 2010)
115. Ghisletta P, Spini D. An introduction to generalized estimating equations and an application to assess selectivity effects in a longitudinal study on very old individuals. Journal of Educational and Behavioral Statistics. 2004;29(4):421-437
116. Hanley JA, Negassa A, Edwardes MD, Forrester JE. Statistical analysis of correlated data using generalized estimating equations: an orientation. American Journal of Epidemiology. 2003;157:364–375
117. Abdulnabi R. Traditional repeated measures analysis versus random coefficients models using PROC MIXED. Paper SP10. Pfizer Inc.,
118. Ahn J. Beyond single equation regression analysis: path analysis and multi-stage regression analysis. American Journal of Pharmaceutical Education. 2002 Spring;66:37–42
119. Carrière I, Bouyer J. Choosing marginal or random-effects models for longitudinal binary responses: application to self-reported disability among older persons. BMC Medical Research Methodology. BMC Med Res Methodol. 2002 Dec 5;2:15. doi:10.1186/1471-2288-2-15
129
120. Kuss O. How to use SAS for logistic regression with correlated Data. Paper 261-27. Institute of Medical Epidemiology, Biostatistics, and Informatics, Halle/Saale, Germany
121. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988 Dec;44(4):1049-60
122. Jaccard J. Interaction effects in logistic regression. Sage publication, 2001
123. Hox J. Multilevel analysis: techniques and applications, Publishers: Lawrence Erlbaum Associates, Inc., 2002
124. UCLA: Academic Technology Services, Statistical Consulting Group. Introduction to sas. This study downloaded their initial sas codes from http://www.ats.ucla.edu/stat/sas (accessed April 22, 2010) and modified the codes for this study purpose.
125. Sun W, Chou CP, Stacy AW, Ma H, Unger J, Gallaher P. SAS and SPSS macros to calculate standardized Cronbach’s alpha using the upper bound of the phi coefficient for dichotomous items, Behavior Research Methods 2007;39(1):71-81
126. Nunnally JC. Psychometric theory, 2nd ed., New York: McGraw-Hill. 1978
127. SAS Institute Inc. SAS/STAT user’s guide, version 8, Cary, NC: SAS Institute Inc., 1999
128. Menard S. Applied logistic regression analysis, the second edition. A Sage Publications, Inc., 2002
129. Armenian HK, Shapiro S. Epidemiology and health services, New York, Oxford University Press, 1998, p.135
130. Campbell DT, Stanley JC. Experimental and quasi-experimental designs for research, Reprinted from Handbook of Research on Teaching, Houghton Mifflin Company, 1963
131. Tom B, Stalker GM. The management of innovation, London: Tavistock, 1961
132. Child J. Organization: contemporary principles and practice. Blackwell Publishing. 2005, p.26
133. Abdolrasulnia M, Menachemi N, Shewchuk RM, Ginter PM, Duncan WJ, Brooks RG. Market effects on electronic health record adoption by physicians. Health Care Management Review. 2008 July/September;33(3):243-252
134. Damanpour F. Organizational innovation: a meta-analysis of effects of determinants and moderators. Academy of Management Journal. 1991;34(3):555-590
135. Earl M, Edwards B, Feeny D. Configuring the IS function in complex organizations, in "information management: the organizational dimension" edited by Earl M. 1996, 201-230, Oxford University Press, Oxford, New York.
130
136. Park YT, Speedie SM. Association of increased admissions with the extent of hospital HIT use. American Medical Informatics Association (AMIA) 2009 Annual Symposium, poster, 2009, Nov. 14-18, Hilton San Francisco, San Francisco, CA
137. Himmelstein DU, Wright A, Woolhandler S. Hospital computing and the costs and quality of care: a national study. the American Journal of Medicine, November 24 (Online version), 2009
138. Theera-Ampornpunt N. Measurement of health information technology adoption: a review of the literature and instrument development, Master of Health Informatics (Master project), University of Minnesota, August 2009
139. Institute of Medicine, Committee on Quality of Health Care in America. To err is human: building a safer health system. Kohn LT, Corrigan JM, Donaldson MS, editors. Washington, DC: National Academy Press; 2000.
140. Bates DW, Cohen M, Leape LL, Overhage JM, Shabot MM, Sheridan T. Reducing the frequency of errors in medicine using information technology. J Am Med Inform Assoc. 2001;8:299-308
131
Appendix 1. Hospital Information Technology Survey (English)
Dear Chief Information Officer or Manager, My name is Young-Taek Park and I am a Ph.D. student in the Division of Health Informatics at the University of Minnesota. Although Health Information Technology (HIT) has been proven to impact decision-making, management processes, quality of care and the bottom line, many hospitals have yet to adopt HIT. To this end, I have been studying the adoption of the Electronic Medical Record (EMR)/Electronic Health Record (EHR) system, use, and interoperability in hospitals. Assuming your participation in this brief survey, upon completion of my study, I can email you the aggregate results of the survey as well as your specific responses. I believe that the study results will extend the knowledge boundary of EMR/EHR systems. The results will also be used for hospital managers and policymakers to establish good decision making and policy alternatives related to the introduction of the EMR/EHR system. The data from this survey will only inform research for my doctoral dissertation. It will only be shared in aggregate form, and will not be used for any other reasons. I cordially request your cooperation in acquiring reliable data. If you are interested in receiving a copy of the executive summary, please contact me directly via email. Your time and participation is truly appreciated. April 1, 2009 Park, Young-Taek Ph.D. Student Division of Health Informatics University of Minnesota If you have any questions, please call me: (Cell) 1-612-481-3760, (Home) 612-331-5266, or (Office) 1-612-625-9431 or please send me an e-mail (Email: [email protected]). This web-based online survey is available at http://www.tc.umn.edu/~park0601
Release: I understand that my participation in this survey is voluntary and that I may exit or quit the survey at anytime without repercussions.
132
1. Questions for follow-up this survey:
Respondent’s name:
Respondent’s telephone number:
E-mail address:
2. Is this hospital part of a multi-hospital system? (1) Yes (2) No 3. If this hospital has any contracts such as group purchasing arrangement with other hospitals, please √ check the first year your hospital started this arrangement.
Before 2003 2003 2004 2005 2006 2007 2008 2009
4. Please check all the following supported by the Hospital Information System in your hospital and please √ check the year it was introduced, if applicable.
Hospital Information System Introduced? Year of introduction
Yes Before2003 2003 2004 2005 2006 2007 2008 2009
Outpatient CPOE*
Inpatient CPOE
Drug management/dispensing
Patient charging processing
Clinical laboratory work
Radiology work
Operating room
Intensive Care Unit
ED information system*
Administrative procedures
* CPOE: Computerized Physician Order Entry , ** ED: emergency department
133
5. An Electronic Health Record (EHR) system is defined as “a system that integrates electronically originated and maintained patient-level clinical health information, derived from multiple sources, into one point of access” (AHA, 2007). An Electronic Medical Record (EMR) system is defined as a system that has “the record of patient health information generated by encounters at one particular provider, which is the electronic replacement of paper charts” (Texas Medical Association, 2006). If your hospital is equipped with either an EMR or an EHR system, please complete the boxes below.
Departments Implemented
Introduction Year
Type of Installation
Fully Partially Not Internal Development Outsourced Purchased
Hospital wards
Outpatient clinics
6. What are the available functions in your hospital’s EMR/EHR system? (Please check √)
Functions Fully Implemented
Partially Implemented
(1) Access to current medical records (observations, orders) (2) Access to medical history (3) Access to patient flow sheets (4) Access to patient demographics (5) Order entry or results review – Lab (6) Order entry or results review – Any radiology report (7) Order entry - real time drug interaction alerts (8) Other clinical alerts (9) Clinical guidelines and pathways (10) Patient access to electronic records (11) Patient support through home-monitoring, self-testing, and interactive patient education
7. Does your hospital’s EMR/EHR system have the ability to transmit and receive these categories of information electronically from other clinics or hospitals (EMR/EHR information exchange)?
Type of information exchanged* Fully exchanged
Partially exchanged No
(1) Patient history (family, patient) (2) Patient summary (conditions, meds, and allergies) (3) Discharge summaries (4) Laboratory information (5) Imaging information (6) Pharmacy information (7) Other
134
8. How does your hospital transfer medical information to other clinics or hospitals located in different areas? (Multiple choice) (1) Paper document delivery or receiving (e.g., patient referral form) (2) Fax (3) E-mail (with attached documents) (4) Internet access to patient records (5) Patient card system such as IC card (6) Health Information Technology (HIT) system-to-system (7) Phone call messages (8) No sharing (9) Other 9. Frequency of the CIO / IT department manager / IT department director’s participation in:
Always Often Sometimes Seldom Never
Before EMR/EHR introduction
The decision to hire new staff?
The decisions on the promotion of any of the professional staff?
Decisions on the adoption of new programs?
Decisions on the adoption of new policies?
Now
The decision to hire new staff?
The decisions on the promotion of any of the professional staff?
Decisions on the adoption of new programs?
Decisions on the adoption of new policies?
10. What are the specialties of your IT Department employees and how many were there before EMR/EHR adoption?
Specialty or major Total No. of IT Department Employees
Before EMR/EHR introduction Now
Physician (Medicine major) Nurse (nursing major) Management/Business/Administration major Medical Technology major Computer science major Etc. * Full time equivalent employees, e.g., physician 1, nurse2, medical technician 5, etc.
135
11. Did your hospital use any of the following standards before EMR/EHR adoption and does your hospital use them now?
Functions Before EMR/EHR Adoption (Yes)
Now (Yes)
(1) Unified Medical Language System (UMLS) (2) Logical Observation Identifiers Names and Codes (LOINC) (3) Health Level Seven (HL7) (4) Digital Imaging and Communications in Medicine (DICOM) (5) SNOMED-CT (6) the National Drug Code (NDC) (6) the National Drug Code (NDC) (7) Nursing Minimum Data Set (NMDS) (8) Nursing Management Minimum Data Set (NMMDS) (9) Omaha System (10) International Classification for Nursing Practice (ICNP) * Systematized Nomenclature of Medicine-Clinical Terms
12. If your hospital’s structure is most like A, then check (√) 1. If your hospital’ structure is most like B, then check (√) 5. Otherwise, check the best one representing how close your hospital is either A or B (3: Middle position between A and B).
A (Hospital Structure) 1 2 3 4 5 B (Hospital Structure)
(1) Hierarchic structure of control, authority, and communication
A network or horizontal structure of control, authority, and communication
(2) Emphasis on formal communication channels (limited access to managerial information)
Emphasis on open communication channels (Not limited access to managerial information)
(3) Strong insistence on a uniform managerial style throughout the business unit
Managers’ operating styles allowed to range freely from the very formal to the very informal
(4) A strong emphasis on holding fast to true and tried management principles despite any changes in business conditions
A strong emphasis on adapting freely to changing circumstances without too much concern for past practice
(5) Strong emphasis on always getting personnel to follow the formally defined procedures
Strong emphasis on getting things done even if this means disregarding formal procedures
(6) Tight formal control of most operations by means of sophisticated control and information systems
Loose, informal control; informal relations and norm of co-operation for getting work done
(7) Strong emphasis on getting line and staff personnel to adhere closely to formal job descriptions
Strong tendency to let the requirements of the situation and the individual’s personality define proper on-job behavior
136
Appendix 2. Hospital Information Technology Survey (Korean language)
(Information Technology: IT) 병원의정보기술 실태조사
, 전산담당최고책임자또는관리자님께
, 저의 이름은 박영택이라고 하며 미네소타 주립대학교 보건정보관리학과 박사 과정에 있는. , 학생있습니다 보건정보기술이 의사결정 및 관리지원 양질의 의료제공 등에 긍정적인 영향을
, . 미침에도 불구하고 많은 병원들이 보건정보기술의 도입을 주저하고 있는 실정입니다 이러한, (EMR)/ (EHR) , , 이유로 저는 전자의무기록 전자건강기록 의 도입 이용 그리고 정보교류 현황에
. , 대한 공부를 해오고 있습니다 현재 병원에 관한 자료를 수집하고 있으며 자료수집에 참여해주시면 귀병원의 현황과 전체병원의 통계에 대한 자료를 이메일로 송부해 드릴 수
있 . 습니다
EMR/EHR 저는 본 연구의 결과가 시스템과 관련하여 기존에 우리가 알고 있는 지식영역을 . EMR/EHR 넗히게 될 것이라고 생각합니다 또한 본연구결과는 병원의 시스템 도입과 관련하여 병원의 관리자와 정부의 정책입안자가 좋은 의사결정과 정책대안을 만드는데 유용하게
. 쓰이리라생각합니다
. 본조사로부터얻은자료는저의박사학위논문을위하여쓰일것입니다 오직요약된형태로써, . 쓰이며 다른 어떠한 이유로는 쓰이지 않을 예정입니다 이 러한 자료수집에 정중하게 도움을
. 요청하는바입니다 만약조사결과에대한요약된자료를받기를원하시면저에게이메을통하여 .연락을주십시요
. 시간을내여참여해주셔서진심으로감사드립니다
2009 4 1년 월 일
박영택 박사과정
보건정보관리학과 미네소타주립대학교
, : ( ) ( ) 1질문이 있으시면 미네소타 주립대학교 박영택 전화 미국 핸드폰 -612-481-3760, ( ) 1집 -612-
331-5266, or ( ) 1학교 -612-625-9431, : 이메일 [email protected] . 로연락을주십시요 (이자료조사는웹싸이트를통하여온라인으로 http://www.tc.umn.edu/~park0601) . 가능합니다
: , 안내문 이 설문에 대한 참여는 자발적인 것이며 어떠한 것에도 영향을 받지 않고 언제든지 설문지 작성을
그만둘 . 수있습니다
137
1. :설문지관리를위한응답자에대한질문
:응답자의직책및이름
:응답자의전화번호
( ):응답자의이메일주소 설문확인을위함
2. (a multi본병원은하나의재단또는법인으로다병원체계 -hospital system) ? 로운영되고있습니까 (1) (2) 예 아니요
3. ( , 같은재단이나법인이아닌다른병원과협력관계를맺고있으면 예 협력병원, 자매병원, 모자병원 등), 형태에 관계없이
맨 처음 협력을 시작한 연도를 √ 해 주십시요?
2003이전 2003 2004 2005 2006 2007 2008 2009
4. , 본 병원의 병원정보화시스템에 의해 지원되는 다음의 모든것에 대하여 체크하여 주시고 가능하면 해당 시스템의
도입년도를√ . 해주십시요
병원정보시스템
도입되었?나요 도입연도
예 2003 이전 2003 2004 2005 2006 2007 2008 2009
외래 처방전달 (OCS)
입원 처방전달 (OCS)
약제업무 및 관리
환자 원무관리
임상병리 검사업무
방사선과의 검사업무
수술실 진료업무
중환자실 진료업무
응급실 진료업무
행정업무
* OCS: Order Communication System. ** ED: emergency department.
138
5. (Electronic Health Record: EHR) 전자건강기록 시스템 “ 은 병원내부 및 외부의 여러 부서에서 전자적으로 생성된 환자에
, ” ( , 2007), 대한 임상정보를 통합 및 유지하여 필요시 모든 것에 신속하게 접속할 수 있는 하나의 시스템 으로 미국병원협회
(Electronic Medical Record: EMR)전자의무기록 시스템 “은 하나의 특정 의사에 의하여 생성된 환자 의 임상정보기록으로
종이의무기록을전자적 기록 으로대체하는 것” (의 Texas Medical Association, 2006) 전자적시스템으로 정의할 수 있습니다.
만약 본 병원에서 EMR 또는 EHR 시스템을 운영하고 있다면, 다음 사항에 대하여 응답하여 주십시요
설치부서
실행형태 도입연도
개발형태
완전 부분적 미실행 내부적개발 외주용역 구입
병 동
외 래
6. EMR/EHR ? 본병원의 시스템에는어떠한기능들이있습니까 (√ 표기를해주십시요)
주요기능 완전한실행 부분적실행
(1) 현재의의무기록에 (Medical records) 대한 접근 (관찰, 처방입력)
(2) 진료기록에 (Medical history) 대한 접근
(3) 환자의 경과기록지에 (Patient flow sheets) 대한 접근
(4) 환자의 기초기록에 (Patient demographics) 대한 접근
(5) 처방입력/결과리뷰 (Order entry/results review):병리실(Labs)
(6) 처방입력/결과리뷰: 방사선기록* (Any radiology report)
(7) 처방입력 – 실시간 약품부작용에 대한 경보 (Real time drug interaction alerts)
(8) 기타 임상적인 기록에 대한 경보 (Other clinical alerts)
(9) 임상가이드라인 및 치료경로 제공 (Clinical guidelines and pathways)
(10) 전자기록에대한환자의접근 (Patient access to electronic records)
(11) 홈모니터링, 자가검진, 대화식 환자교육을 통한 환자지원
* 또는영상의학기록 7. EMR/EHR 본 병원의 시스템은 다음과 같은 카테고리의 정보를 다른 병원이나 의원과 전자적으로 주고 받을 수 있는
(EMR/EHR )? 기능이있습니까 정보교환
*주고받을수있는정보의종류 전체적 교류 부분적 교류 없음
(1) (Pati환자의히스토리 ent history) (가족력, 환자력) (family, patient)
(2) 환자기록 (Patient summary) 요약지 (현상태, 약품처방 내력, 알러지)
(3) 퇴원요약지 (Discharge summaries)
(4) 임상병리정보 (Laboratory information)
(5) 영상이미지정보 (Imaging information)
(6) 환자에대한처방약품 정보 (Drug information)
(7) 기타
139
8. ( )?본병원은어떻게의료정보를다른지역의의원이나병원들에전송을하고있습니까 복수응답가능 (1) ( . ) 종이차트전달및접수 예 환자의뢰서 (2) 팩스밀리머신 (3) 이메일및자료첨부기능 (4) 환자기록에대한인터넷접속 (5) IC card 와같은환자카드시스템 (6) 의료정보공유(Health Information Technology: HIT) 시스템의 이용 (7) 전화메시지 (8) 정보공유 없음 (9) 기타 9. 본원의 전산담당 최고책임자 (실장, 부장, 또는 과장) 의 병원내의 의사결정에 대한 참여정도:
항상 종종 때때로 가끔 전혀
EMR/EHR 도입이전
신규직원 채용을 위한 의사결정
기존직원의 승진에 대한 의사결정
새로운운영프로그램 도입의 의사결정
새로운 규정채택의 의사결정
현재
신규직원 채용을 위한 의사결정
기존직원의 승진에 대한 의사결정
새로운 운영프로그램 도입의 의사결정
새로운 규정채택의 의사결정
10. EMR/EHR , 도입전의전산부서의대략적인전문인력규모는어떠했으며 현재는어떠합니까 ?
( )전문인력 전공 전산부서의정규인력규모
EMR/EHR 도입이전 현재
( )의사 의학전공 명 명
( )간호사 간호전공 명 명
/ /경영관리행정 명 명
(Medical technologies)의학기술관련 명 명
컴퓨터관련전공 명 명
기타 명 명
* 예., 의사 1 명, 간호사2 명, 의료기사 5 명, 5기타 명 11. EMR/본병원의 EHR ? 도입전표준화시스템은어떠했으며현재는어떤것을사용하고있습니까
기 능 EMR/ EHR 도입이전 사용
(예) 현재 사용
(예)
(1) Unified Medical Language System (UMLS)
(2) Logical Observation Identifiers Names and Codes (LOINC)
(3) Health Level Seven (HL7)
(4) Digital Imaging and Communications in Medicine (DICOM)
(5) SNOMED-CT*
(6) 한국의약품표준코드(KD코드, Korea Drug Code)
(7) Nursing Minimum Data Set (NMDS)
(8) Nursing Management Minimum Data Set (NMMDS)
(9) Omaha System
(10) International Classification for Nursing Practice (ICNP)
* Systematized Nomenclature of Medicine-Clinical Terms
140
12. 병원 조직 구조가 A 에 가까우면 1에 √ 체크해 주시고, B에 가까우면 5에√ 체크를 해 주십시요. 그렇지 않으면 A 나 B 를 잘 반영하는 것에√ 체크하여 주십시요 (3은 A와 B의 중간 정도에 위치합니다).
A (병원조직 구조) 1 2 3 4 5 B (병원조직 구조)
(1) 권한위임, 교류, 관리체계가 수직적 권한위임, 교류, 관리체계가 수평적 또는 네트웍 중심적
(2) 공식적인 의사소통 채널 (제한적 경영정보 접근)
개방적 의사소통 채널 (자유로운 경영정보 접근 가능)
(3) 조직부서의 동일한 관리스타일 중시 조직부서의 다양한 관리스타일 인정
(4) 상황변화와 상관없이 검증된 관리원칙 고수
과거원칙에관계없이상황변화에 민감하게대처하는것강조
(5) 담 당자를통한공식적 업무처리절차의 강조
공식적인 절차보다 문제해결의 성과강조
(6) 잘 정립된 통제관리 장치에 의한 엄격한 공식통제
느슨한 비공식적 통제, 규범이나 비공식적 협력 관계에 의한 통제
(7) 직무기술서에 명시된 업무결재라인 강조
상황이나개인특성에따른의사결정인정
- ! 설문에참여해주셔서감사합니다 -
141
Appendix 3. IRB Letter
142
Appendix 4. University of Minnesota home page for the survey
143
Appendix 5. Survey introduction and password section
144
Appendix 6. Survey main page
145
Appendix 6. Survey main page (continued)
146
Appendix 6. Survey main page (continued)
147
Appendix 6. Survey main page (continued)
148
Appendix 6. Survey main page (continued)